Innate Immune Memory and Transcriptional Profiling of Microglia
Heng, Yang
DOI:
10.33612/diss.151944032
IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from it. Please check the document version below.
Document Version
Publisher's PDF, also known as Version of record
Publication date: 2021
Link to publication in University of Groningen/UMCG research database
Citation for published version (APA):
Heng, Y. (2021). Innate Immune Memory and Transcriptional Profiling of Microglia. University of Groningen. https://doi.org/10.33612/diss.151944032
Copyright
Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).
Take-down policy
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.
Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the number of authors shown on this cover page is limited to 10 maximum.
General introduction and outline of the thesis
Outline of general introduction
Microglia are the tissue-resident macrophages of the central nervous system (CNS), which are specifically tailored to their function to maintain CNS homeostasis. Perturbations of brain homeostasis that result from infection or disease can trigger
microglia activation 1. In the past decade, thanks to the advances in technologies such
as RNA sequencing, epigenetic profiling, cellular fate mapping, imaging, and proteomics, research on microglia is flourishing. In this General Introduction, relevant literature is summarized and discussed, with a particular focus on the application of the approaches and technologies mentioned above. Section 1 discusses the origin of the microglia and the similarities and differences with other tissue-resident macrophages. Section 2 introduces how microglia are maintained under physiological or pathological conditions during adulthood. Studies on longevity and turnover of microglia are also summarized. In Section 3, different microglia depletion and repopulation strategies are compared. Section 4 summarizes the epigenetic, transcriptional and surface protein signature of microglia in comparison to other myeloid cell types. Section 5 introduces region-specific and sex-specific microglia heterogeneity. Section 6 introduces innate immune memory in microglia. In Section 7 the diversity in microglia phenotypes during ageing and in CNS diseases is summarized.
1. Origin of microglia
Microglia have a similar but distinct ontogeny compared to other tissue-resident macrophages. Microglia and other tissue-resident macrophages initially derive from
early erythro-myeloid precursors (EMPs) during the first wave of hematopoiesis 2. In
the mouse, the first wave starts at the yolk sac posterior plate mesoderm around E7.0. At E7.5, early EMPs are generated which further differentiate into yolk sac
macrophages without passing through monocytic intermediates 3. The yolk sac
macrophages next colonize the entire embryo, including the brain and spinal cord and differentiate into microglia in the CNS and other types of macrophages in other tissues
3. Generation of early EMPs requires colony-stimulating factor 1 receptor (CSF1R) 4, and
1
required for development of hematopoietic stem cells (HSCs) and all CD11bhi
monocytes and macrophages later during development 6. Although microglia and other tissue-resident macrophages can derive from early EMPs, unlike microglia that remain of early EMP origin, early EMP-derived macrophages in many other embryonic tissues are replaced by fetal monocyte-derived macrophages during the second wave of hematopoiesis 7. At E8.25, the second wave of hematopoiesis (definitive hematopoiesis) starts in the yolk sac, generating late EMPs 7. Different from early EMPs, late EMPs do not require CSF1R expression but are MYB-dependent 7. Most of these late EMPs (some can also contribute to yolk sac macrophages) migrate to the fetal liver (FL) at E9.5, where they differentiate into myeloid progenitors 7. At E12.5, myeloid progenitors start to differentiate into FL monocytes. At this time point, only
yolk sac macrophages (CD45+CD11bloF4/80hiLy6C-), but not FL monocytes
(CD45+CD11bhiF4/80loLy6C+) are present in the liver, skin, kidney, lung, and brain
rudiments 7,8. At E14.5, large numbers of FL monocytes invade all tissues except the brain, which is probably due to the formation of the blood brain barrier (BBB) that started around E13.5 7. These invaded FL monocytes develop into tissue macrophages, diluting the number of original yolk sac macrophages 3,9. Thus, in the end, late EMPs are the main precursors for most tissue-resident macrophages 7. In addition, tissue resident macrophages can partially derive from HSCs generated from the third and final wave of hematopoiesis 2,3. In contrast to the observations mentioned
above, using a KitMercreMer fate mapping mouse line, it has been shown that most tissue
resident macrophages are derived from HSC, with the exception of microglia and some
Langerhans cells, which are yolk sac-derived 10. Although the origin of other tissue
resident macrophages is still under debate, there is a general consensus that microglia are derived from yolk sac EMP.
2. Maintenance of the microglia population
Under homeostatic conditions, tissue-resident macrophages self-maintain their
population with minimal contribution of circulating monocytes 11. Nevertheless, some
tissue macrophages like alveolar, kidney, and heart macrophages show declining self-renewal ability with age and are progressively replaced by monocyte-derived
macrophages 12. As tissue-resident macrophages in the CNS, microglia exclusively rely on self-renewal under physiological conditions during adulthood.
2.1. Local self-renewal of microglia
In response to pathogens or other CNS damage, microglia alter their numbers, morphology, surface marker and gene expression, a process that is termed microgliosis
13. For decades, it was believed that bone marrow (BM)-derived blood-circulating
monocytes contributed to the expansion of microglia in response to CNS damage 14.
Using parabiotic mice, it has been shown that maintenance and local expansion of microglia exclusively relies on self-renewal both under physiological and certain
pathological conditions 15,16. These pathological conditions include amyotrophic lateral
sclerosis (ALS), facial nerve axotomy, and experimental autoimmune encephalomyelitis
(EAE) 15,16. However, with preconditioning of the brain by irradiation, circulating
monocytes (Ly6ChiCCR2+) were observed at lesioned brain areas and differentiated into
microglia-like cells in cuprizone-induced demyelination and facial nerve axotomy
mouse models 17. These cells were indistinguishable from microglia in terms of Iba1
and F4/80 expression 17. Nevertheless, these monocyte-derived macrophages did not
contribute to the microglia pool after inflammation had been resolved 16,17. Using
Ccr2rfp:Cx3cr1gfp mice and transcriptional profiling, Yamasaki et al. successfully
distinguished monocytes-derived macrophages and microglia in EAE mice, and showed that monocyte-derived macrophages have distinct morphological features and relatively high phagocytic and inflammatory transcriptional profiles compared to
resident microglia 18. These observations suggest that under normal conditions,
microglia are a self-sustained cell population without peripheral contribution during adulthood.
2.2. Longevity and turnover of microglia
In recent years, our understanding of microglia lifespan, proliferation and turnover has been greatly increased due to the application of fate mapping and imaging techniques, providing us more insights into how these cells are dynamically maintained in the brain. It has been shown that microglia proliferation and apoptosis are temporally and
1 nucleotide analog that is incorporated into the DNA during S-phase, Askew et al found
that 0.69% of the microglia cells are proliferating (here defined as BrdU+ population) in
the mouse brain, and the DG region has the highest percentage of proliferating
microglia cells (1.25%) among various brain regions (Table 1) 19.
Table 1. Mouse and human microglia lifespan, proliferation rate, proliferating population and turnover time
Study Species Method Lifespan Proliferation
rate (per day) Turnover time Proliferating population
Askew et
al. 19 mouse BrdU label
96 days 0.69% on average; 1.25% in DG; 0.4% to 0.7% in CTX, CA1–2, TH, and OB Eco-SFFV-RV 0.4% in CTX; 0.5% in STR in vivo imaging 0.79% in OB
human Ki67 2% in CTX
Tay et al. 20 mouse EdU and BrdU label 0.39% in OB 0.22% in HPP 0.17% in CBM 0.16% in HTH 0.09% in MB 0.07% in CTX 8 m in OB 15 m in HPP 20 m in CBM 21 m in HTH 38 m in MD 41 m in CTX Ki67 0.27% in CTX; 0.41% in pooled microglia from OB, HPP, CBM Füger et al. 21 mouse in vivo imaging 29 m in CTX 0.07% in CTX Réu et al. 22 human 14C analyses;
IdU 4.2 years 0.08% by
14C;
0.14% by IdU in CTX
Abbreviations: BrdU, bromo-2'-deoxyuridine; CA, cornu ammonis; CBM, cerebellum; CTX, cortex; DG, dentate gyrus; Eco-SFFV-RV, Eco-SFFV γ-retroviral vector; EdU, ethynyl-2’-deoxyuridine; HPP, hippocampus; HTH, hypothalamus; IdU, 5-iodo-2′-deoxyuridine; OB, olfactory bulb; MB, midbrain.
Based on an estimated cell cycle duration of 32 h, they calculated that the entire
microglia population turns over once every 96 days in mice 19. By Ki67 staining, both
young (20-35 years old) and aged (58-76 years old) humans were shown to have a similar percentage of proliferating microglia cells (2%), which is higher than was observed in mice. But the different proliferation rates can also result from the fact that Ki67 expression is not directly comparable to BrdU incorporation as Ki67 is also
expressed in other cell cycle stages besides the S-phase 19.
Using a Cx3cr1creER:R26RConfetti mouse line, a multicolor fluorescence fate mapping
cortex 20. However, in response to injury (facial nerve axotomy), this stochastic
proliferation shifts to clonal microglial expansion 20. Using EdU and BrdU labeling, it has
been shown that microglia proliferation rate varies among different brain regions
(Table 1) 21. Based on the measured proliferation rates, microglia were estimated to
turn over within 41 months in the cortex, 15 months in the hippocampus and 8 months
in the olfactory bulb 20. By genetical labeling of approximately 2% of microglia, Füger
et al. individually and temporally tracked microglia cells in vivo using multiphoton
microscopy over 15 months 21. By analyzing the disappearance of microglia, they
estimated that microglia lifespan is around 29 and 22 months in young (4-month) and
adult (10-month) mice, respectively 21. In addition, the number of newly generated
microglia (13%) was similar to the number of disappeared microglia, and the newly generated microglia were still present in the cortex over the total 6-month imaging
period 21. These data suggest that the proliferation rate in the cortex is around 0.07%
per day (13% divided by 6 months), which is quite similar to previous findings by BrdU
labeling (Table 1) 20. Also, the comparable percentages of generated and lost microglia
during the imaging period suggests that microglia self-renewal is maintained by
coupled proliferation and apoptosis, as shown by Askew et al. 19. By measuring 14C
content in the genomic DNA of human cortical microglia, Réu et al. calculated that the average human microglia lifespan is 4.2 years with a proliferation rate of 0.08% per day
(28% per year) in the cortex (Table 1) 22.
These studies only have analyzed the lifetime of microglia during steady state, and more in-depth studies are needed to confirm these findings and determine microglia dynamics in disease conditions.
3. Microglia depletion and repopulation systems
In addition to studying microglia turnover under steady state conditions, microglia ablation systems provide more insight into how these cells are maintained and replenished in the CNS under extreme conditions. In general, two strategies are used to deplete microglia in vivo during adulthood: pharmacologic and genetic depletion (Table 2).
1
Table 2. Comparison of selected microglia depletion systems
System Treatment Depletion
efficiency Side effects Repopulation time Origin of repopulated cells
CSF1R inhibitor PLX3397 23; PLX5622 24 90% (7 d), 99% (21 d) 23 upregulated GFAP 23 7 d 25 non-microglia Nestin+ progenitor cells 23; remaining microglia 26; Mac2+ remaining microglia 27 Cx3cr1CreER:
R26iDTR tamoxifen + DT 99% (1 d after DT) 28; 80% (3
d after first DT) 29 impaired motor learning ability, cytokine storm, astrogliosis 29 14 d after first DT 29 remaining microglia 29; remaining microglia and monocyte-derived macrophages 30 Cd11b-HSVTK irradiation + GCV i.p. injection 30% (4 wks) 31 astrogliosis, myelotoxicity 31,32 i.c.v. injection GCV pumps 95% (2 wks), 97% (4 wks) 31 2 wks 33 infiltrating monocytes 33 Cx3cr1CreER: Csf1rFlox tamoxifen diet 25% (1 to 4 wks) 34 Monocyte-derived macrophages 34 Abbreviations: DT, diphtheria toxin; GCV, ganciclovir; GFAP, glial fibrillary acidic protein. 3.1. Pharmacologic microglia depletion
Treatment of mice with a CSF1R inhibitor (PLX3397) results in rapid ablation of
microglia in the brain. With 7 days of treatment, 90% of microglia were eliminated 23. Within 7 days after withdrawal of the inhibitor, a fast and complete repopulation of the CNS with microglia was observed 23. It has been shown that the repopulation capacity of microglia is limited, as microglia in mice subjected to multiple rounds of elimination (7 days treatment) and repopulation (7 days recovery) failed to repopulate the brain within 7 days 25. However, when the recovery time was extended to 28 days, microglia still repopulated the entire brain 25. Surprisingly, a 99% loss of microglia did not affect
cognitive and motor function as measured by a series of tests including the Barnes
maze, contextual fear conditioning, elevated plus maze, open field and rotarod 23. In
addition, at the timepoint analyzed, no cytokine storm was induced by microglia
ablation in the brain 26. Similarly, in a study by Rojo et al., Csf1r gene enhancer deletion resulted in ablation of microglia and other resident macrophages in the skin, kidney, heart and peritoneum. Nevertheless, these mice were healthy and fertile, without the growth, neurological or developmental abnormalities reported in Csf1r -/- mice and rats 35. At repopulation day 60, brain transcriptomes are comparable to controls (only 3 differentially expressed genes: Gabra6, Cbln3 and Fat2), and repopulated and resident
microglia respond similarly to a peripheral LPS challenge, suggesting that original and
repopulated microglia are transcriptionally and functionally similar 26.
For the origin of these repopulated microglia, Elmore et al. first demonstrated that
repopulated microglia are derived from non-microglial Nestin+ progenitor cells in the
CNS without contribution from the periphery 23. In contrast, Huang et al. reported that
the repopulated microglia are solely derived through proliferation of the residual
microglia, and not from Nestin+ progenitors 26. In a more recent study, Zhan et al.
demonstrated that the microglia escaping the action of the CSF1R inhibitor are Mac2 (Galectin-3)-positive and highly enriched genes associated with microglia early
development 27. These results indicate that, after pharmacologic depletion by a CSF1R
inhibitor, the repopulated microglia are derived from the remaining microglia population without contribution from cells from the periphery.
Recently, manipulating the microglia pool by depletion and repopulation using CSF1R inhibitors has been shown to be a promising approach to treat CNS disorders. In aged 3xTg-AD mice (15 months), partial depletion (30%) of microglia by a low dose of CSF1R inhibitor for 3 months prevented microglia association with plaques and improved learning and memory as measured in a Morris water maze and a novel place recognition
test 36. In an inducible neuronal lesion mouse model (CaM/Tet-DTA), both microglia
depletion, and depletion followed by repopulation could resolve neuronal lesion induced neuroinflammation and behavioral deficits in an elevated plus maze and a
Morris water maze test 24,37. In a traumatic brain injury (TBI) model, depletion followed
by repopulation attenuated spatial learning deficits and stimulated neurogenesis after
the injury 38. Nevertheless, microglia depletion alone did not improve the outcome. In
the maternal immune activation (MIA) model, MIA induced upregulation of genes related to cellular neuritogenic pathways in microglia, synaptic dysfunction in frontal
cortex, repetitive behavior and social deficits in the offspring 39. Microglia depletion and
repopulation attenuated these synaptic, neurophysiological, and behavioral
abnormalities 39
3.2. Genetic microglia depletion
In 2005, Heppner et al. generated mice carrying a CD11b-HSVTK transgene in which the thymidine kinase of herpes simplex virus (encoded by HSVTK) is expressed under
1
control of the CD11b promoter 32. HSVTK converts the antiviral nucleotide analog
prodrug ganciclovir (GCV) to a monophosphorylated form, which is further
transformed into a toxic triphosphate 32
. After systemic GCV administration, in CD11b-HSVTK mice, circulating CD11b+ cells were substantially reduced after 3 days and
almost entirely ablated after 5–6 days 32. To overcome this GCV-mediated
myelotoxicity, chimeric mice were transplanted with wild type bone marrow. Oral GCV administration for 4 weeks resulted in a ~30% decrease in microglia numbers in the
neocortex in bone marrow chimeric APPPS1 mice 31. To circumvent bone marrow
transplantation, the CD11b-HSVTK system has been further developed by direct
administration of GCV intracerebroventricularly (i.c.v.) in APPPS1 mice 31. After 2
weeks of i.c.v. GCV treatment, 95% of Iba1+ cells were depleted. After 4 weeks, 97% of
microglia were depleted, and astrogliosis occurred 31. A follow-up study from the same
group showed that microglia could repopulate the brain within 2 weeks after removal of GCV treatment. However, the repopulated microglia were two times higher in number and morphologically different from resident microglia (with shorter,
asymmetrical processes and enlarged soma) 33. In addition, these cells expressed high
levels of CD45 and CCR2, suggestive of a peripheral origin 33.
In 2013, Parkhurst et al. used Cx3cr1CreER transgenic mice to drive diphtheria toxin
receptor (DTR) expression in CX3CR1-expressing cells, resulting in the
Cx3cr1CreER:R26iDTR system where microglia can be selectively depleted upon diphtheria
toxin (DT) administration 28. One day after the last DT administration (1 μg, i.p. for three
consecutive days), 99% microglia were depleted in the brain. At 7 days after DT administration, the number of microglia remained significantly lower (85% reduction), but was already higher than 1 day after treatment, demonstrating that microglia
repopulation had already started 28. Indeed, using the same system, Bruttger et al.
observed efficient microglia depletion, with 80% depletion of microglia at day 3 after the first DT injection, which was followed by a fast repopulation that started before day
7 and was complete by day 14 29. Microglia depletion by the Cx3cr1CreER:R26iDTR system
reduced motor learning ability 28, and induced production of cytokines and chemokines
(such as IL-1β, TNF-α, CXCL2, and CXCL9), as well as astrogliosis in the brain 29. It has
originated from the remaining cells in the Cx3cr1CreER:R26iDTR system 29. With irradiation
or chemotherapy-induced myeloablation, Lund et al. showed that peripheral Ly6Chi
monocytes can infiltrate the brain and become monocyte-derived macrophages in the
Cx3cr1CreER:R26iDTR system30. These monocyte-derived macrophages can acquire some
key features of microglia but still are transcriptionally and functionally distinct from
CNS resident microglia even 12 weeks after depletion 30. Similarly, Shemer et al. showed
that after BM transplantation, the BM-derived macrophages in the brain have distinct transcriptomes and epigenetic landscapes and respond differently to a peripheral LPS
challenge 40.
Using Cx3cr1CreER:Csf1rFlox mice to specifically delete Csf1r in Cx3cr1-expressing cells,
Cronk et al. found that Csf1r-deletion led to a partial microglia loss (25% reduction in number of microglia), which resulted in engraftment of peripherally derived macrophages. But the engrafted macrophages are transcriptionally distinct from
resident microglia 34. Comparisons of the above-mentioned depletion systems are
summarized in Table 2. More depletion systems that act during the embryonic stage are
reviewed elsewhere 41.
4. Microglia identity
In the past decade, phenotyping of microglia has greatly benefitted from advances in technologies like whole-genome transcriptomic and epigenomic profiling, unbiased proteomics, and cytometry by time of flight (CyTOF). In this section, some of the important findings on microglia identity are summarized, with a focus on microglia epigenetic, transcriptional and surface protein signatures. In addition, similarities and differences with CNS-associated macrophages (CAM) are also discussed here.
4.1. Microglia epigenetic signature
Cells in a multicellular organism have the same genome; however, they are structurally and functionally distinct owing to cell type specific epigenetic regulation of gene
expression 42,43. DNA is folded around histone proteins to form nucleosomes, and these
nucleosomes can be opened to expose their DNA sites to enable the initiation of
transcription 44. Whether or not a chromosomal region is open is determined by
1
histone modifications 45,46. Among them, histone modifications are most studied in
microglia. Histone modifications are chemical groups such as methyl, acetyl or phosphate that are covalently bound to specific amino acids usually in the histone
amino terminal tails 45. The levels and types of modifications on histones play important
roles in transcriptional regulation 45. For example, mono-methylation of histone 3 at
lysine 4 (H3K4me1) is correlated with gene activation in microglia 47. Other commonly
studied histone modifications are provided in Box 1
In 2014, Lavin et al. generated genome-wide epigenetic profiles of histone modifications H3K4me1, H3K4me2, H3K4me3, and H3K27ac across monocytes,
neutrophils and seven macrophage populations, including microglia 47. These myeloid
cell types share a large percentage of H3K4me3-marked promoters (8,861 of 10,806
promoters, 82%). However, less than 2% of H3K4me1-marked enhancers are shared between these seven populations, indicating cell identity is mainly determined by the
active enhancer repertoire in these myeloid cells 47. Compared to monocytes and other
macrophages (spleen red pulp macrophages, liver Kupffer cells, lung macrophages, peritoneal cavity macrophages, and colonic large and ileal small intestinal macrophages), microglia showed specific enrichment of H3K4me1 at enhancers of
Sall1, Siglech, Fcrls, and Sparc (summarized in Table 3) 47. Since H3K4me1-marked
enhancers can be either active or poised, they further analyzed the distribution of H3K27ac in marked enhancers to identify active enhancers. H3K4me1-marked enhancers of Cx3cr1 were shared by all macrophages, but they were only active
in microglia and intestinal macrophages 47. Previous studies showed that PU.1-binding
sites are co-enriched for motifs of several lineage-determining transcription factors
(TFs) such as C/EBPα and AP-1, which determines macrophage cell identity and cell-specific response 48-50. By combining H3K4me1 ChIP-seq with ATAC-seq, Lavin et al.
identified Mef2 as another likely lineage-determining TF in microglia 47. In another
study, Gosselin et al. examined H3K4me2 and H3K27ac histone modifications in
Box 1. Commonly used chromatin marks and their functional association in microglia
H3K4me1 histone H3 lysine 4 mono-methylation mark active or poised enhancers
H3K4me2 histone H3 lysine 4 di-methylation mark active or poised promoters and enhancers
H3K4me3 histone H3 lysine 4 tri-methylation mark active promoters
H3K27ac histone H3 lysine 27 acetylation mark active enhancers >> promoters
microglia, large peritoneal macrophages (LPM, MHCIIlo), small peritoneal macrophages
(SPM, MHCIIhi), thioglycollate-elicited peritoneal macrophages (TGEM) and
bone-marrow-derived macrophages (BMDM) 51. Similarly, when comparing H3K4me2
patterns in microglia to LMP, they also found more differential H3K4me2-marked regions at potential enhancers (27%; total 36,607) than at promoters (3%; total 7,937)
51. Sall3 enhancers were exclusively marked by H3K4me2 and H3K27ac (active
enhancer) in microglia compared to other macrophages 51. Super-enhancers (SE) are
enhancers where both the median length and the levels of associated mediator protein
are at least an order of magnitude larger/higher than of the typical enhancer 52. Some
SEs are thought to control the expression of genes that define cell identity 43. When
compared with LPM and TGM, they found that 45% of the SEs (total 576, identified by H3K27ac ChIP-seq) were specific to microglia, such as the SEs of Gpr56 and Cx3cr1
(Table 3) 51. For human microglia, Gosselin et al. performed motif enrichment analysis of ATAC-seq-defined open chromatin regions associated with H3K4me2. They found that PU.1, CTCF, IRF, RUNX, MEF2, CEBP, AP-1 and SMAD motifs were highly enriched in enhancers both in human and mouse microglia 53. In 2019, Nott et al. used ATAC-seq, H3K27ac and H3K4me3 ChIP-seq to obtain epigenetic profiles from nuclei of different brain cell types including microglia, neurons, astrocytes and oligodendrocytes. Similar to previous
findings in mice 47,51, brain cell identity in humans is also mainly controlled by
enhancers 54. Further, using proximity ligation-assisted ChIP-seq (PLAC-seq), a method
to detect long-range chromatin interactions. Nott et al. identified 219,509 significant unique interactions between enhancers and H3K4me3-marked promoters across
microglia, neurons and oligodendrocytes 54. A substantial proportion of cell
type-specific gene promoters are PLAC-linked to SEs 54, some of the examples for microglia
are listed in Table 3. Remarkably, several SEs harbored genome-wide association study (GWAS) disease-risk variants and were also linked to cell type-specific genes
promoters, suggesting that some GWAS variants act on SE to affect gene expression 54.
For microglia, sporadic Alzheimer’s disease (AD) variants were largely found at microglia enhancers (Table 3, with asterisk), suggesting microglia could be more
1
Table 3. Microglia-specific enhancers
Species Comparison Method Examples of target genes of microglia specific enhancers
mouse monocytes, neutrophils and macrophages from six other tissues ATAC-seq; H3K4me1 and H3K27ac ChIP-seq Sall1, Siglech, Fcrls, Sparc, Mef2 LPM, SPM, TGEM,
BMDM H3K4me2 and H3K27ac ChIP-seq Sall3, Gpr56 (compared with LPM and TGM), Cx3cr1 (compared with LPM and TGM)
human neurons, oligodendrocytes PLAC-seq (H3K4me3)
TFs: CEBPA, EGR2, ETS2, FLI1, FOS, IRF8, MEF2C, RUNX1, RUNX2, SPI1, STAT6 Receptors: CD14, CSF1R, CX3CR1, ITGAM*, ITGAX*, P2RY12 Others: AIF1, ALOX5, APOE*, B2M, BIN1*, CASS4*, IL6, INPP5D*, PICALM*, TGFB1
Abbreviations: BMDM, bone-marrow-derived macrophage; LPM, large peritoneal macrophage; SPM, small peritoneal
macrophage; TF, transcription factor; TGEM, thioglycollate-elicited peritoneal macrophage. Genes with asterisk represent
AD-risk variants identified by GWAS. This table is adapted from Nott et al., 2019 54, and combined data from Lavin et al. 47
and Gosselin et al 51.
4.2. Microglia transcriptional signature
Chromatin modification at enhancers has been shown to correlate with cell-type
specific gene expression 55, suggesting the existence of a set of enhancers that
determines cell-type specific gene expression patterns. But how these enhancers are selected and promote transcription of cell-type specific genes is unknown. Recently, it was proposed that the selection of cell type-specific enhancers is based on the binding site within the enhancers and the interaction of lineage-determining transcription
factors (LDTF) and signal-dependent transcription factors (SDTF) 56. In microglia, PU.1 is an important LDTF, which interacts with SMADs, MEF2 and other TFs, to regulate the transcription of microglia specific genes 51,57. The gene expression profile of mouse and human microglia was first identified using bulk population samples, e.g. a large number of purified microglia in one sample, often from a relatively large brain region or complete hemispheres. For mouse microglia, the first core microglia signature was generated in 2012 using microarrays in the
Immunological Genome (ImmGen) project 58. Based on this study, Chiu et al. compared
microglia microarray data with data from 22 other myeloid cell types collected by the ImmGen project. 99 genes were identified that were 5-fold or more enriched in
microglia relative to other myeloid immune cells 59. Furthermore, they also compared
spinal cord microglia RNA-seq data with RNA-seq data obtained from astroglia, motor neurons, and whole spinal cord, yielding 288 genes enriched in microglia. The overlap between two data sets identified 29 highly specific markers for microglia, including
Olfm3, Tmem119 and Siglech 59. By direct RNA sequencing of sorted microglia and whole brain samples, Hickman et al. identified a cluster of genes responsible for microglia sensing functions, referred to as the microglia sensome. Comparison with peritoneal macrophages identified 626 differentially expressed transcripts and the top 25 most highly expressed microglia transcripts include several sensome genes: P2ry12, P2ry13,
Tmem119, Gpr34, Siglech, Trem2 and Cx3cr1 60. These microglia signatures were
confirmed in two studies that addressed the transcriptomic and epigenetic differences
between microglia and other tissue-resident macrophages 47,51. By gene profiling and
quantitative mass spectrometry analysis, Butovsky et al. identified 1,572 genes and 455
proteins enriched in microglia compared to CD11b+Ly6C+ spleen-derived monocytes 61.
Based on these two datasets, a Nanostring chip was designed to further investigate the
differences between microglia and F4/80+ CD11b+ macrophages derived from
peripheral organs, 239 genes were specifically expressed by microglia. When compared to other CNS cells (astrocytes, oligodendrocytes and neurons), 106 genes were microglia specific. P2ry12, Fcrls, Tmem119, Olfml3, Hexb and Tgfbr1 were identified as unique microglial genes; PU.1 as a microglia-specific transcription factor; and three microglia-specific microRNAs were identified (125b-5p, 342-3p and
miR-99a) 61. Importantly, these microglia signature genes were not expressed by newborn
microglia (P1), cultured primary microglia (P1-2), microglia cell lines (N9, BV2) or by
embryonic stem cell-derived microglia 61. Using the microglia marker Tmem119, a
mouse microglia gene expression profile during development and 1 day after an LPS
challenge was generated 62. During development, 37 of 100 top microglia-enriched
genes were consistently upregulated from E17 to P60. Again, a homeostatic microglia
core signature was identified, and LPS induced a typical inflammatory gene profile 62.
Collectively, these studies led to the identification of a homeostatic mouse microglia core gene expression signature, which includes Sall1, Hexb, Fcrls, Gpr43, Cx3cr1, Tmem119, Trem2, P2ry12, Mertk, Pros1, and Siglech genes that are abundantly expressed in microglia compared to other brain or myeloid cells. Details on isolation methods, tissues used and which kind of comparison were used to identify mouse microglia signature genes in these studies are summarized in Table 4.
1 Ta bl e 4 . M ou se a n d h u m an m ic ro gl ia tr an sc ri p to m es id en ti fi ed by bu lk p op u la ti on s equ en ci n g St u d y Mi cr o gl ia is o la ti o n me th o d Ti ss u e u se d De te ct io n me th o d Co m p ar is o n Si gn at u re g en es Re p re se n ta ti ve ge n es Ga ut ie r et al ., 20 12 58 En zy m at ic di ss oc ia ti on w it h Li be ra se II I, P er co ll-gr ad ie nt s ep ar at io n, an d FA CS s or te d as CD 11 b +CD 45 loF4 /8 0 lo 6-we ek m al e C5 7B L/ 6 mo us e br ai n Mi cr oa rr ay Co m p ar ed w it h s pl ee n re d p ul p m ac ro ph age s (F 4/ 80 hi B2 20 ne g CD 11 c hi MH C I hi); p er it on ea l ma cr oph ag es (C D 11 5 hi F 4/ 80 hiMH CI I ne g); lu ng ma cr oph ag es (S ig le c-F +CD 11 c +MH C II lo). Ge ne s up re gu la te d in m ic ro gl ia b y 5-fo ld o r mo re r el at iv e to th ei r ex pr es si on in th e th re e ot he r m ac rop h ag e po pu la ti on s 65 ge ne s Cx 3Cr 1, Si gl ec h, Tm em 11 9, Sa ll1 , He xb Ch iu e t al ., 2013 59 Me ch an ic al di ss oc ia ti on , P er co ll gr ad ie nt s ep ar at io n, CD 11 b + ma gn et ic be ad s pu ri fi ca ti on SO D 1 G9 3A, no n-Tg , SO D 1 WT an d (L P S) in je ct ed mo us e sp in al c or d RN A -se q Co m p ar ed w it h R N A -se q da ta o bt ai n ed fr om as tr og lia, m ot or n eu ro ns a nd w ho le sp in al co rd ( 5-fo ld in cr ea se ; q < 0 .0 5) 288 ge ne s Ol fm 3, Si gl ec h, Tm em 11 9, Ga l3 st 4, Cs m d3 , Sl co 2b 1 Co m p ar ed w it h m ic ro ar ra y da ta fr om 2 2 ot he r m ye loi d ce ll ty pe c ol le ct ed by Im m un ol og ic al G en om e P roj ec t ( G au ti er e t al ., 2012; 5 -fo ld in cr ea se , q < 0 .0 5) 99 ge ne s Th e ov er la pp in g ge ne s b et w ee n th e tw o d at a se ts m en ti on ed a b ov e 29 ge ne s Hi ck m an et al ., 2013 60 Di ss oc ia te d by G en tl e Ma cs w it h en zy m es (d is pa se , c ol la ge na se III) , P er col l g ra di en t se pa ra ti on , a nd F A CS so rt ed b ase d on CD 11 b an d CD 45 5-mo nt h-ol d C5 7B L/ 6 mo us e br ai n Di re ct R N A - se q Fi rs t i de nt if ie d 1, 2 99 s en so m e ge ne s. N ex t, co m pa re d w it h w ho le b ra in , t op 1 00 s en so m e ge ne s hi gh ly e nr ic he d in m ic ro gl ia b as ed o n E va lu e w er e se le ct ed (E =C M M R [m ic ro gl ia ]/ CM M R [b ra in ]) 100 se ns om e ge ne s P2 ry 12 , P2 ry 13 , Tm em 11 9, Gp r3 4, Si gl ec h, Cx 3c r1 , Tr em 2 Co m p ar ed th e to p 10 % o f t ra nsc ri pt s w it h th e hi ghe st e xp re ss io n in m ic ro gl ia w it h tho se in pe ri to ne al m ac ro ph ag es ( CD 11 b +CD 45 +) 626 ge ne s un iq u el y ex pr es sed in m ic ro gl ia Am on g th e 25 m os t h ig hl y ex pr es se d tr an sc ri pts th at w er e al so u ni qu el y ex pr es se d in m ic ro gl ia o ve r ma cr oph ag es ( p < 0. 00 001, lo g2 fo ld c ha ng e > 4) , m ic ro gli a se ns om e ge ne s w er e id en ti fi ed 7 se ns om e ge n es o ut o f 25 ge ne s
St u d y Mi cr o gl ia is o la ti o n me th o d Ti ss u e u se d De te ct io n me th o d Co m p ar is o n Si gn at u re g en es Re p re se n ta ti ve ge n es Bu to vs ky et al ., 2014 61 Si ng le ce ll su sp en si on , Pe rc ol l g ra di en t se pa ra ti on , a nd F A CS so rt ed b ase d on CD 11 b + CD 45 lo C5 7B L6 mo us e br ai n Mi cr oa rr ay Co m p ar ed w it h CD 1 1b +Ly 6C + sp le en -de ri ve d mo no cy te s 1, 572 ge ne s Fc rl s, P2 ry 12 , Me rt k, Pr os 1 Ma ss sp ec tr om et ry Co m p ar ed w it h CD 1 1b +Ly 6C hi an d CD 11 b +Ly 6C lo sp le en -de ri ve d m on oc yt es (m as s sp ec tr om et ry ; g re at er th an 2 -fo ld di ff er en ce ). 455 pr ot ein s (74 of th em un iq ue ly e xp re ss ed in mi cr og lia ) P2 ry 12 , Lg m n, Tp pp , Bi n1 , Rg s1 0 Mi cr oa rr ay w er e ge ne ra te d ba se d on a b ov e me nt io n ed ma ss sp ec tr om et ry an d ar ray d at a Co m p ar ed w it h 1 0 ty pe s of im m un e ce lls , 8 ty pe s of F 4/ 80 + CD 11 b + or ga n m ac ro ph ag es 239 ge ne s Fc rl s, C1 qa , P2 ry 12 , Pr os 1, Me rt k, Ga s6 Co m p ar ed t o ot he r CN S ce lls ( as tr oc yt es , ol ig od en dr oc yt es a nd n eu ron s) 106 ge ne s Fc rl s, Ol fm l3 , Tm em 11 9, P2 ry 12 , He xb , Tg fb r1 Mi cr oa rr ay co nt ai ne d 6 00 mi cr oR N A s Co m p ar ed t o 6 ty pe s of o rg an m ac ro ph age s an d 11 ty pes o f i m m un e cel ls 8 m ic ro R N As h ig hl y ex pr es sed mi R -125b -5p , m iR -342 -3p , m iR -99 a Be nn et t e t al ., 20 16 62 Me ch an ic al di ss oc ia ti on , m ye li n re m ov al b y be ad s, a nd FA CS s or te d b as ed o n Tm em 11 9 + (E 17 a nd pa rt o f P 7 sa mpl es we re b as ed o n CD 45 lo CD 11 b+ s or ti ng) E1 7, P 7, P 14 , P2 1 an d P6 0 mo us e br ai n an d LP S in je ct ed mo us e br ai n Pa ir ed -en d R N A -se q Fi rs t, to p 10 0 m ic ro gl ia e nr ic he d g en es o ve r no n-mi cr og lia C N S my el oi d ce lls ( T me m1 19 -CD 11 b + CD 45 hi) fr om P 60 (> 16 -fo ld ) we re id en tif ie d. N ex t, co ns is te nt m ic ro gl ia e nr ic he d ge ne s du ri ng th e de ve lo pm en t we re id en tif ie d. 37 of 100 to p m ic ro gl ia -en ri ch ed g en es ar e up re gu la te d fr om E 17 to P6 0 Ol fm l3 , Fc rl s, Sl c2 a5 , Lt c4 s, Cx 3Cr 1, Se lp lg , P2 ry 12 , Cc r5 , Pl xd c2
1 St u d y Mi cr o gl ia is o la ti o n me th o d Ti ss u e u se d De te ct io n me th o d Co m p ar is o n Si gn at u re g en es Re p re se n ta ti ve ge n es Go ss el in et al ., 2017 63 Me ch an ic al di ss oc ia ti on , P er co ll gr ad ie nt s ep ar at io n, an d FA CS s or te d as CD 11 b +CD 45 LoCD 64 +C X3 CR 1 Hi li ve c ells , ex cl ud in g m os t ac ti vat ed c el ls w it h mo de ra te to h ig h le ve ls o f C D 45 19 hu m an br ai n ti ss ue s re se ct ed fro m tr ea tm en t of ep ilep sy , br ai n tu m or s, or a cu te is ch em ia RN A -se q Mi cr og lia l g en e si gn at ur e ge ne s w er e id en tif ie d w it h a cu to ff o f 1 0-fo ld in cr ea se d ex pr es si on r el at iv e to c or tex ti ss ue (F D R < 0. 05) 881 ge ne s C3 , CS F1 R , SP P1 , CX 3CR 1, P2 RY 12 , C1 Q B , C1 Q A Fi rs t, di ff er en ti al ly e xp re ss ed ge ne s lis t i n ne ur od eg en er at iv e and b eh av io ra l d is or d er s de ri ve d fr om m ic ro ar ra y or R N A -se q of in ta ct ti ss ue fr om 4 6 pu bl ic ly a va ila bl e da ta s ets we re o bt ai n ed . S ec on d, o ve rl ap pi ng g en es be tw ee n 8 81 m ic ro gli a si gn at ur e ge ne s w it h tho se d if fe re nt ia lly re gul at ed g en es w ere id en tif ie d. Fo r ex am pl e, 9 7 mi cr og lia g en es w er e po si ti ve ly c or re la te d wi th B ra ak s ta ge o f Al zh eim er 's d is ea se (AD ) in p re fr on ta l c or te x ti ss ue C1 Q A , C1 Q C, IR A K 3, SP P1 , SL C1 A5 , SL C7 A7 , TN FR SF 10 B , TN FR SF 1B Ba se d on p re vi ou sl y id en ti fi ed r is k al le le s A D , Pa rk in so n' s di se ase ( PD ), m ul ti pl e sc le ro si s (M S) an d sc hi zo ph re ni a (S cz ) ri sk a lle le s 64, ma ny o f t he se g en es w er e pr ef er en ti al ly ex pr es sed in m ic ro gl ia co m par ed t o co rt ex ti ss ue AD , f or e xa m pl e, 2 8 of 48 AD g ene s w er e hi gh er in mi cr og lia TR E M 2, SO RL 1, IN PP 5D , ME F2 C, CD 33 Ga la tr o et al ., 20 17 65. Me ch an ic al di ss oc ia ti on , P er co ll gr ad ie nt s ep ar at io n, an d FA CS s or te d as DA P I ne gCD 11 b hi ghCD 4 5 in t ev en ts Po st m or te m ri ght p ar ie ta l co rt ex fr om do no r wi th ou t ap par en t ne ur op at ho lo gi ca l ab no rm al it ies RN A -se q Mi cr og lia l g en e si gn at ur e ge ne s w er e id en tif ie d w it h a cu to ff lo g fo ld c ha ng e > 3 an d ad ju st ed p < 0. 00 1 co m pa re d to c or te x ti ss ue 1, 297 ge ne s CX 3CR 1, CS FR 1, C1 Q A -C , CL E CL 1, CI IT A, IT G A M , IC A M -1 , CD 33 , SI GL EC 5/ 7-12 /1 4 Us in g ag e as a q ua nti ta ti ve v ar ia bl e, m icr og lia si gn at ur e ge ne e xp re ssi on w as ex am in ed in do no rs r an ge d be tw ee n 34 a nd 1 02 ye ar s 212 ge ne s in cr ea se d an d 360 ge ne s de cr ea se d in ex pr es si on d ur in g ag ei ng IT G A L, TL N 1, PF N 1, VA SP , P2 RY 12 , IL 6R , TL R 10 , IC A M 3, RO B O 2, SE M A3 C
St u d y Mi cr o gl ia is o la ti o n me th o d Ti ss u e u se d De te ct io n me th o d Co m p ar is o n Si gn at u re g en es Re p re se n ta ti ve ge n es Gr ab er t e t al ., 20 16 66 Me ch an ic al di ss oc ia ti on , P er co ll gr ad ie nt s ep ar at io n, pu ri fi ed b y an ti -CD 11 b m ic ro be ad s Hi pp oc am pu s , c or te x, ce re be llu m , an d st ri at um is ol at ed fr om mi ce b ra in Mi cr oa rr ay Ge ne s di ff er en ti al ly e xp re ss ed b y br ai n re gi on ( p < 0. 05 w it h F D R c or re ct io n ). 2, 527 ge ne s Ge ne c oe xp re ss io n an al ys is o f t he r eg io n-sp ec if ic mi cr og li al ph en ot ype s by B io La yo ut Ex pr es s Ge ne s en ri ch ed in ce re be llu m m icr og li a we re im m u ne -re la te d Cl ec 4e , Cl ec 7a , St at 1, St at 4, Ir f7 , Oa sl 1, H2 -D1 , H2 -Aa , CD 74 Ge ne s en ri ch ed in ce re be llu m a nd hi pp oc am pus m ic ro gl ia re la te d to e ne rg y pr od uc ti on sy st em Pf kp , Ca t, So d1 , So d2 , Md h1 , Pp ar g, Nd uf a1 , At p5 a1 , Co x5 b Ay at a et al ., 20 18 67 Tr an sl at in g ri bo so m e af fi ni ty p ur if ic at io n (T R AP ). M ec ha ni ca l di ss oc ia ti on , c el l l ys is , an d R N A w as p ur if ied by a nt i-GF P b ea ds Mi cr og lia RN A is ol at ed fr om ce re be llu m an d st ri at um fr om T R A P mi ce Mi cr og lia -sp ec if ic TR A P-se qu en ci ng ; ( al so co nfi rm ed b y si ng le -nu cl ei RN A -se q) Di ff er en ti al e xp re ss io n a na ly si s by DE Se q2 so ft w ar e 297 ce re be lla r m ic ro gl ia en ri ch ed g en es (a ss oc ia te d w it h c el l cl ea ra nce fu nct io ns ) Ax l, H2 -Aa , Ap oe , Mr c1 , Cd 74 , Ly z2 , Li lr b4 , Co le c1 2 733 st ri at um m ic ro gl ia en ri ch ed g en es (a ss oc ia te d w it h m at ur e mi cr og lia -sp ec if ic ho m eo st at ic su rv ei lla nc e) Sl c2 a5 , Ec sc r, Fs cn 1, Sa ll3 , Ar hg ap 30 , Ac ap 2, Fc rl s, Ma fb , Hh ex Gu ne yk ay a et al ., 20 18 68 Me ch an ic al di ss oc ia ti on , P er co ll gr ad ie nt s ep ar at io n, an d pu ri fi ed b y an ti -CD 11 b m ag ne ti c be ad s Ma le a nd fe m al e m ou se hi pp oc am pus an d co rt ex RN A -se q Ge ne s di ff er en ti al ly e xp re ss ed b et w ee n ma le s an d fe ma le s in h ippo ca mpu s an d co rt ex ( ad ju st ed p < 0. 01, lo g2 fo ld c ha n ge > 0. 5 or < − 0. 5) 1, 109 ge ne s un iq ue ly di ff er en ti al ly e xp re ss ed in h ip po ca m pu s, 55 ge ne s in h ip po ca m pu s. 46 ge ne s w er e di ff er en ti al ly e xp re ss ed in b ot h re gi on Ir ak 1, Tm em 50 b, Tm em 33 , Tm em 30 a, At p1 1c , At p6 ap 2, Kd m 5d Va n de r Po el e t a l., 2019 69 Me ch an ic al ly di ss oc ia ti on , en zy m at ic d ig es ti on wi th c ol la ge n as e, Pe rc ol l g ra di en t se pa ra ti on , n eg at iv e se le ct io n w it h an ti -CD 15 a nd p os it iv e se le ct io n w it h an ti -CD 11 b b ea ds Hu m an b ra in co rt ica l G M , co rp us ca llo su m W M RN A -se q Ge ne s di ff er en ti al ly e xp re ss ed b et w ee n GM an d W M in c on tr ol an d M S do no rs ( fo ld ch an ge > 2 o r < −2 , p < 0. 05 w it h FD R co rr ect io n) 453 ge ne s in c on tr ol , 124 ge ne s in M S do n ors GM e nr ic he d ge ne s: CCL 2, TN FR SF 25 ; W M en ri ch ed g en es : CX CR 4, AC K R 1, GP N MB , NU PR 1 Ab br ev ia ti on s: AD , Al zh ei m er 's d is ea se ; F AC S, fl uo re sc en ce -ac ti vat ed c el l s or ti ng ; G M , g rey m at ter ; L PS , l ip op ol ys ac ch ar id e; M S, m ul ti pl e sc ler os is ; P D , P ar ki ns on 's d is eas e; S cz , sc hi zo ph re ni a; T R A P, tr an sl at in g ri bo so m e af fi ni ty p ur if ic at io n; W M , w hi te m at te r
1
In two studies published in 2017 63,65, the human microglia transcriptome was reported.
Gosselin et al. expression-profiled microglia isolated from surgically resected brain tissue of epilepsy, brain tumor, or acute ischemia patients, hence without postmortem delay (PMD). Microglia were isolated by Percoll gradient centrifugation and
fluorescence-activated cell sorting (FACS) of live-CD11b+CD45lowCD64+CX3CR1high
cells, while excluding most activated cells with moderate to high levels of CD45. The 30 most abundant transcripts in microglia across different patients were related to known microglia properties and functions like ramification and motility (P2RY12 and CX3CR1), synaptic remodeling (complement components C3 and C1QA-C), and antigen
presentation (HLA-DRA and HLA-B) 63. Comparison of the microglia transcriptome to
the transcriptome of cortical brain tissue used to isolate microglia, resulted in the
detection of 881 microglia-enriched (10-fold increased) genes 63.
In the study by Galatro et al., microglia were isolated from postmortem CNS tissues from donors without apparent neuropathological abnormalities. Microglia were
isolated by mechanical dissociation, Percoll gradient centrifugation, and sorted as live-CD11bhighCD45int cells. Compared to corresponding cortical tissue, 1,297 human
microglia signature genes were detected (logFC>3 enrichment). Gene ontology (GO) analysis showed many significantly enriched terms and genes associated with the biological functions of microglia, such as immune signaling and modulation (CD74, CSFR1, C1QA-C), pathogen and self-recognition (MyD88, CLECL1, CIITA), and cell adhesion and motility (ITGAM, CX3CR1, ICAM-1). Also, human microglia expressed many SIGLECs (CD33, SIGLEC5/7-12/14), showing their important role in maintenance
of CNS homeostasis 65.
In both studies, the human and mouse microglia transcriptomes were compared. Gosselin et al. showed that human and mouse microglia are very similar, 13,253 of 15,768 orthologous genes pairs expressed within a 4-fold range. At a cutoff of 10-fold difference, they identified 400 human microglia enriched orthologous genes and 293 mouse microglia enriched orthologous genes. Human microglia are characterized by a higher expression of regulators of the complement system (C2, C3, VSIG4, SERPING1) and genes involved in brain structure development (SYNDIG1, GLDN, CTTNBP2 and ROBO3). Genes more abundantly expressed in mouse microglia included Hexb (related
to microglia lysosome function), Sparc (related to microglia proliferation and
structure), and Sall3 (microglia specific transcription factor) 63,70-72.
In the study of Galatro et al., the human microglia transcriptome was also compared
with mouse microglia transcriptomes 66,73. An extensive overlap with human microglial
data was observed, however, noteworthy differences were also discovered. Human microglia-specific genes were shown to be involved in immune pathways, example
genes are GNLY, CD58, APOBEC3C, CLECL1, CD89 and CARD8 65. Most notably, when
comparing age-associated changes in microglia gene expression between humans and
mice 65,66, a suprisingly low overlap was detected. Most genes with an aging-associated
change in expression in humans were associated with the actin cytoskeleton 65. Detailed
information on isolation methods, tissues used and comparisons used to identify human microglia signature genes in these studies is provided in Table 4
4.3. Microglia surface protein signature and comparison with CNS-associated macrophages (CAMs)
Microglia express classic macrophage markers like F4/80, CD11b, CD45 (at a lower level than macrophage), CD64, CSF1R, CD200R1, CX3CR1, tyrosine–protein kinase Mer
(MERTK) and the ionized calcium-binding adaptor molecule 1 (IBA1) 74,75. Through
transcriptional profiling as discussed above, more surface markers were identified such as P2Y purinoceptor 12 (P2RY12), transmembrane protein 119 (TMEM119), Fc receptor-like S, scavenger receptor (FCRLS) and ectonucleoside triphosphate diphosphohydrolase 1 (ENTPD1; also known as CD39) and sialic acid binding Ig-like
lectin H (SIGLECH) 74.
Compared to microglia which reside in the CNS parenchyma, CAMs or border-associated macrophages (BAMs) reside in non-parenchymal interfaces such as the perivascular space, the meninges, and the choroid plexus (Figure 1). Similar to microglia, meningeal and perivascular CAMs also originate from the yolk sac and maintain their population by self-renewal without contribution from peripheral
myeloid cells 76,77. CAMs share common surface markers with microglia such as CX3CR1,
CD11b, CSF1R, F4/80, CD64, MERTK and IBA1 77, making it challenging to distinguish
them when their anatomical origin is not known 75. Still, some surface proteins are
1
believed to be CD45lo MHCIIlo, while CAMs are CD45hi MHCIIhi 76. However, a recent
study shows that one CAM subset (CD38+MHCIIloCCR2lo) also expresses low levels of
CD45 76,78, indicating CD45 is not sufficient to distinguish these two populations. In
terms of surface proteins, CAMs specifically express CD38, CD163, CD206 and
lymphatic vessel endothelial hyaluronan receptor 1 (LYVE1) compared to microglia
77-79 (Figure 1). On the transcriptional level, Tmem119, P2ry12, Hexb, Sparc, and Olfml3 are
defined as microglia-enriched genes, whereas CAMs highly express genes such as
Ms4a7, Mrc1, Pf4, Stab1, and Cbr2 77,80,81 (Figure 1). Very recently, Masuda et al. have
generated a HexbtdTomato mouse line in which microglia are specifically labelled, but not
CAMS, offering a valuable tool to study brain macrophages 82.
Figure 1. Resident tissue macrophage populations at CNS interfaces and their signature genes and surface markers compared to microglia. CNS-associated macrophages (CAMs) are localized at the CNS
interfaces (a). The best understood CAM populations are the leptomeningeal macrophages, perivascular macrophage and the choroid plexus macrophages. Recently, through single-cell RNA-seq and single-cell mass and fluorescence cytometry, several signature genes and surface markers for CAMs and microglia have been
identified (b). *, a CAM subset (CD38+MHCIIlo CCR2lo) also expresses low levels of CD45.
5. Microglia heterogeneity
Microglia are highly plastic cells and their morphology, phenotype, and immune
response display region-dependent heterogeneity 83,84. A detailed study of the basal
ganglia region revealed region-specific phenotypes of microglia and this microglial
transcriptional level, Grabert et al. were first to demonstrate gene expression differences between microglia from the cerebral cortex, hippocampus, cerebellum, and striatum, which became more pronounced during aging. Cerebellar and hippocampal microglia exhibited a more immune-vigilant state, with higher expression of the genes
Camp and H2-Ab1 66. Using a microglia-specific translating ribosome affinity
purification (TRAP) approach, it was determined that cerebellar microglia displayed a
more pronounced cell clearance phenotype 67
. In contrast, using bulk population RNA-seq, Li et al. detected very limited transcriptomic heterogeneity between Tmem119+
FACS sorted microglia isolated from the brain regions mentioned above 86.
Regional heterogeneity was also demonstrated in the human CNS. Microglia were isolated from grey matter (GM; occipital cortex) and white matter (WM; corpus callosum) from the postmortem CNS of control and multiple sclerosis (MS) donors. Between WM and GM microglia, 453 differentially expressed genes (logFC>2) were detected in samples from control donors, and 124 genes in MS donor-derived samples. Genes highly expressed in control GM microglia were related to “cytokine-mediated signaling”, such as TNFRSF25 and CCL2; WM microglia were enriched for genes involved
in “chemotaxis” and “inflammatory response” (CXCR4, ACKR1, GPNMB, NUPR1) 69.
Besides regional differences, sex-dependent heterogeneity in microglia gene expression was also reported. Transcriptomic profiles were generated of microglia from male and female mouse hippocampus and cortex. Male microglia displayed a higher capacity to present antigens and increased responsiveness to purinergic stimuli
68. Expression profiling of microglia isolated from male and female mice revealed that
the gene expression program in male microglia was delayed 87. Microglia maturation
during development is shaped by microbiome-derived short chain fatty acids 88, and the
effect of the microbiome on microglia differentiation is sexually dimorphic 89.
Perturbation of the microbiome had more profound effects in male embryos and female adults. These studies show that the gut microbiome is important for microglia development and maturation. Table 4 summarizes details on isolation methods, brain regions used and which kind of comparisons were used to identify microglia heterogeneity in these studies.
1
6. Microglia and innate immune memory
In 2011, Netea and colleagues proposed the concept of trained immunity—that innate immune cells also can exhibit immunological memory of past insults, challenging the
classic dichotomy between innate and adaptive immunity 90. Trained immunity refers
to the increase in responsiveness of an innate immune cell to a second infectious
challenge, following the preconditioning by a primary stimulus 90. Notably, this
memory-like response is independent of T and B lymphocytes 91,92, and is not specific
for a particular pathogen 93,94. Trained immunity is supported by the observation of
memory-like response in plants, invertebrates, and innate immune cells in vertebrates
90. As an important experimental evidence in vertebrates, in 2012, the Netea group
showed that after a first C. albicans infection or Bacille Calmette-Guérin (BCG) vaccination, monocytes mounted a heightened response to a second C. albicans
infection, thereby improving the survival rate of mice upon reinfection 91,92. Trained
immunity in monocytes is mediated by epigenetic reprogramming 95-97 as well as a
metabolic shift from oxidative phosphorylation to glycolysis after the first challenge
98-100.
Microglia priming refers to the exaggerated inflammatory response of microglia to a second stimulus compared to naïve microglia, regardless of whether the second
stimulus arises from the CNS or the periphery 1. In general, primed microglia already
have an increased inflammatory state at baseline, and the second stimulus is
superimposed on the ongoing microglia activation 101,102. Microglia priming is observed
in various pathological conditions such as ageing 103, accelerated ageing due to DNA
damage (ERCC1-deficient mice) 104, stress 105, offspring subjected to maternal
inflammation in utero 106, and neurodegenerative diseases such as AD and Parkinson's
disease (PD) 1. Nevertheless, so far, the mechanisms underlying microglia priming are
still unclear.
Thus, both microglia priming and trained immunity are functional definitions, describing the enhanced responsiveness of a cell to the second challenge, and a second challenge is required to reveal the functional consequence of the first stimulus. In terms of trained immunity, the first stimulus is acute, and the second stimulus is generally
chronic, which normally involves chronic neuroinflammation and/or neurodegeneration. And the second stimulus is superimposed on the ongoing chronic
inflammation or neurodegeneration 1,101,109. Although single systemic infection with
live Salmonella typhimurium or traumatic brain injury could also induce microglia priming, also in these cases persistent neuroinflammation was observed after the
infection or injury 110,111. Recently, trained immunity has also been observed in microglia in vitro 112 and in vivo 113. Given that both microglia priming and trained immunity describe the heightened response to a sequential stimulus, Neher and Cunningham proposed to use integrated nomenclature to describe microglia innate immune memory (Figure 2). Figure 2. Integrated nomenclature for microglia innate immune memory. The first stimulus can either be acute (red curve) or chronic (red dotted line), and it can induce persistent molecular changes in microglia. a) If microglia show an enhanced responsiveness to a second stimulus, the first stimulus (I) is called priming stimulus, and the enhanced response is called microglia training. This is regardless of whether the second stimulus (II) is superimposed onto an ongoing inflammation (green dotted curve) or applied after a delay (green curve). b) Conversely, if microglia show a reduced responsiveness to a second stimulus, the first stimulus (I) is called desensitizing stimulus, and the reduced response is called microglia tolerance. This
figure is adapted from Neher and Cunningham, 2019 114.
The first stimulus can either be acute or chronic. If microglia show enhanced responsiveness to a second stimulus, the first stimulus is called priming stimulus. And this priming stimulus induces an immune training response in microglia. In contrast, if microglia show reduced responsiveness to as second stimulus, then, the first stimulus
1 is called desensitizing stimulus. And this desensitizing stimulus induces an immune tolerance response in microglia. This is regardless of whether the second stimulus is
superimposed to an ongoing inflammation or applied after a delay 114. Thus, both
microglia priming and trained immunity can be viewed as immune training in
microglia. But to be consistent with the terminology in our previous publications 102,104, we still used the term ‘microglia priming’ in Chapter 3. In Chapter 4, we adopted the new terms to describe the microglia innate immune responses. 7. Microglia phenotype diversity in ageing and CNS diseases The concept of M1 and M2 polarization was acquired from exposing macrophages to purified stimuli in vitro 115. M1 stimuli can induce pro-inflammatory responses, and M2 stimuli can induce ant-inflammatory responses 115. And this concept was previously also used to define microglia activation. However, as a cell-type with a distinct origin, identity and maintenance compared to monocyte-derived macrophages, microglia have been shown to have diverse profiles in response to ageing, infection and CNS diseases, challenging the implication of this concept in microglia activation 116. Recently, several
transcriptomics studies revealed spatio-temporal diversity of microglia gene
expression in ageing and diseases (Figure 3) 117.
The first disease-associated microglia transcriptomic study was published in 2015 by our group. We identified a common microglia gene expression signature shared
between accelerated aging mice (Ercc1Δ/ko), an AD mouse model, an amyotrophic lateral
sclerosis (ALS) mouse model, and naturally aged mice 102. This profile is characterized
by the upregulation of genes associated with immune, phagosome and antigen
presentation pathways and the downregulation of homeostatic microglia genes 102. In
2017, using single-cell RNA-seq, Keren-Shaul et al. identified a cluster of
disease-associated microglia (DAM) in 5XFAD AD mouse model 118. Similarly, DAMs are
characterized by the upregulation of genes such as Apoe, Trem2 and Tyrobp, which are associated with lipid metabolism and phagocytosis. These genes are also induced in
microglia from aging and CNS disease as reported previously 102. In addition, in these
DAMs, genes associated with homeostatic microglia, such as P2ry12 and Tmem119, were downregulated. While the single-cell RNA-seq study was performed in a mouse
model for AD, the expression of some DAM signature genes was confirmed by immunostaining in human AD brain tissue, where these DAMs spatially associated with sites of AD pathology 118. Figure 3. Microglia phenotype diversity in diseases. In various diseases, microglia adapt a wide range of phenotypes as described in the literature 118-122. However, a common pattern of the transcriptional changes is observed in different diseases, which is characterized by downregulation of some homeostatic genes (green arrow) and upregulation of some disease-associated genes (red arrow). DAM, disease-associated microglia; MGnD, microglial neurodegenerative phenotype; ARM, activated response microglia; IRM, injury-responsive microglia; LPC, lysolecithin.
In another study, a common microglia neurodegenerative phenotype (MGnD) was
identified in mouse models of ALS (SOD1G93A), AD (APP-PS1), and MS (EAE) 119. Similar
to DAMs, MGnD is also characterized by the downregulation of homeostatic microglia genes (P2ry12, Tmem119, Gpr34 and Olfml3) and increased expression of inflammatory genes (Apoe, Spp1, Itgax, Axl, Clec7a, Ccl2 and Csf1). Importantly, they further found the
MGnD signature was dependent on the TREM2-APOE pathway 119. In an App knockin