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.
Chapter G
Summary and general discussion
190
Summary
Microglia play an important role in the maintenance of CNS homeostasis. In the past decade, our insights in microglia biology has been extensively expanded thanks to advances in imaging, RNA sequencing, epigenetic profiling, and cellular fate mapping techniques. Chapter 1 provides a general introduction to this thesis. A selection of recent studies using these techniques regarding microglia origin, maintenance, identity, innate immune responses, and phenotype diversities in disease and steady states are discussed. In Chapter 2, CNS ageing phenotypes in selected mouse ageing models, with a special focus on neurons, astrocytes, oligodendrocytes and microglia are reviewed. Microglia can be chronically activated by the local environment, resulting in an enhanced response to a superimposed inflammatory insult, a phenomenon called microglia priming 1-3. We previously observed microglia priming in constitutive
Ercc1Δ/ko mice 1. In addition, in response to the Ercc1-deficiency, microglia showed a
hypertrophic morphology (thickened processes and a larger soma size) and increased proliferation 1. However, the effect of intrinsic DNA repair deficiency by Ercc1 deletion in microglia is unknown. In Chapter 3, Ercc1 was specifically deleted from Cx3cr1-expressing cells, and changes in microglia were determined at different time points after tamoxifen treatment using the Cx3cr1creERT2:Ercc1ko/loxP system. At 2 months after
tamoxifen treatment, an extensive loss of microglia was observed in various brain regions (cortex, dentate gyrus, and cornus ammonis). In response to the microglia loss, a subset of remaining microglia significantly increased in size and process length. This reduction in cell number and changes in microglia morphology persisted until 12 months after tamoxifen treatment. During this process, Ercc1-deficient microglia (Ercc1ko/rec) were gradually lost in tamoxifen treated Cx3cr1creERT2:Ercc1ko/loxP mice and
replaced by non-deficient Ercc1ko/loxP microglia. Notably, Ercc1-deficient microglia
(Ercc1ko/rec) and non-deficient Ercc1ko/loxP microglia both proliferated similarly as
suggested by comparable Ki67 expression. The replacement of Ercc1-deficient microglia was most likely caused by increased apoptosis of these cells as indicated by gene expression analysis. Gradually, microglia numbers and morphology returned to
7 control levels at 22 months after tamoxifen treatment. Furthermore, we showed that microglia-specific deletion of Ercc1 resulted in a transient ageing phenotype which is characterized by a very different pattern of gene expression from the aged and disease-associated microglia expression signature reported previously by our group 3. In addition to the local environment, past immune experience can also affect microglia immune responses. The capacity of microglia to develop innate immune memory has been demonstrated using LPS as a systemic stimulus in vivo 4,5. In Chapter 4, we assessed whether systemic administration of β-glucan, a fungal cell wall component, could affect microglia innate immune memory. Mice were challenged by i.p. injection of β-glucan, and the acute response of microglia as well as the effect of β-glucan preconditioning on how microglia respond to a second LPS challenge were investigated. We included LPS as a positive control since this TLR4-agonist is well-studied in microglia both in vivo and in vitro. Microglia exhibited a classical inflammatory response to LPS, evidenced by a significant upregulation of Tnf, Il6, Il1β, Ccl2, Ccl3 and Csf1 at 3 h after injection, and obvious morphological changes (increased soma size and retracted processes) at 1 and 2 days after injection. Compared to LPS, β-glucan did not induce expression of these cytokines nor morphological changes in microglia at 3 h, 1 day and 2 days after injection. Nevertheless, the β-glucan receptor gene, Dectin-1 (Cd369, Clec7a), was significantly downregulated at 3 h after injection, suggesting that a peripheral β-glucan challenge activates microglia. β-glucan preconditioning induced immune training in microglia 2 days later, reflected by enhanced responses of Tnf, Il-1β, Il6, and Csf1 expression to a second LPS challenge. However, this training phenotype was no longer apparent at 7 and 14 days after the first challenge. Notably, with a 2-day interval between two challenges, LPS preconditioning also induced immune training in microglia. But the immune training phenotypes induced by β-glucan and LPS were different, indicated by similar but not identical cytokine responses to the second LPS challenge. Both LPS and β-glucan induced immune tolerance in the periphery with a 2-day interval between first and second stimuli, suggesting the immune training observed in microglia was caused by microglia-intrinsic changes.
192
Transcriptional profiling of microglia from 10 different mammalian species revealed that microglia express a conserved core set of signature genes 6. However, compared to microglia from other mammals (mouse, macaque, marmoset, hamster, and sheep), human microglia are more heterogeneous, as identified by single-cell RNA-seq 6. In addition, human microglia more abundantly express genes associated with the complement pathways, phagocytosis, and susceptibility to neurodegeneration compared to rodent microglia 6. Human microglia alter their transcriptional profiles once they were transferred into an in vitro culture environment as early as 6 h after transfer 7. These findings suggest the limitations of investigating microglia in an in vitro culture environment. Thus, the study of freshly isolated microglia from human brain tissue is still irreplaceable. However, due to the ethical and other practical reasons, the access to fresh healthy human brain tissues is limited. Recently, single-nucleus RNA-seq was developed as an alternative approach which can successfully recapitulate single-cell transcriptomes from frozen tissues 8,9. This enables the use of archived postmortem human brain specimens to study microglia in neurodegenerative diseases and neurological disorders.
In Chapter 5, we generated matched microglia nuclear and cellular transcriptomes from fresh mouse and human brain tissues by bulk and single-cell/nucleus RNA-seq. Transcriptomes from freshly isolated mouse and human microglia nuclei were shown to be quite similar to their fresh cellular counterparts. In addition, the transcriptional response of mouse microglia to a peripheral LPS challenge was well-conserved at the nuclear level. Moreover, human microglia nuclei isolated from frozen brain tissues contained all the subpopulations identified in fresh nuclei/cellular samples from the same donors with a quite small variation in cluster ratios. These results indicate that the transcriptome of microglia nuclei from fresh and frozen tissues closely resembles the gene expression profile of freshly isolated cells.
In Chapter 6, microglia were FACS-isolated from mice with increasing postmortem intervals: 0, 4, 6, 12 and 24 h. The numbers of viable microglia (DAPIneg CD11bhigh CD45int Ly-6Cneg) significantly decreased with increasing PMD, but high-quality RNA could still be isolated from samples even with 12 h PMD. RNA-seq results showed that
7 only 50 genes out of 15,554 detected genes were differentially expressed between different PMD time points. These genes were related to mitochondrial, ribosomal and protein-binding functions. Next, we examined the expression of the human homologs of these PMD-related mouse microglia genes in our previously generated transcriptomic data 10. Of the 50 PMD-associated mouse genes, 31 had human homologs. We found that expression of MT-CYTB and MT-ND1 was positively correlated with PMD. For the other 29 homologs, their expression tended to be negatively correlated with PMD, but not significantly. These results indicate that PMD up to 12 h has very limited influence on microglia transcriptomes. Collectively, the data in this thesis led to the following conclusions (Figure 1) relating to the research questions formulated in the final section of the General Introduction.
194
7 Fi gu re 1 . T h e m ai n c on cl u si on s of th e ex p er im en ta l c h ap te rs in th is th es is .
Discussion and future perspectives
Microglia priming as microglia immune training
Recently, trained immunity was observed in microglia both in vivo 11 and in vitro 5. Given that microglia priming and trained immunity both describe the heightened response to a sequential stimulus, Neher and Cunningham proposed to use an integrated nomenclature to describe microglia innate immune memory. The first stimulus could either be acute such as LPS i.p. injection, or chronic such as neurodegeneration, neuroinflammation or other surrounding degenerative changes 12. In response to a second stimulus, the first priming stimulus results in an immune training response (enhanced responsiveness), and a first desensitizing stimulus results in an immune tolerance response (reduced responsiveness). This is regardless of whether the second stimulus is superimposed on an ongoing inflammation or occurs after a delay 12. Thus, both microglia priming and ‘trained immunity’ can be viewed as immune training in microglia 13.
Besides this functional definition of microglia priming, in 2015, our group used a transcriptomic approach to identify a common microglia priming gene expression signature shared in mouse models of natural aging, accelerated ageing (Ercc1Δ/ko mice)
and neurodegenerative diseases (APP and SOD1 transgenic mice) 3. In this study, microglia priming was used to describe a pre-activated state of cells induced by a degenerative environment compared to naïve microglia 3. Importantly, Ercc1Δ/ko mice
microglia, which adopt a microglia priming gene expression signature also showed enhanced responsiveness to a peripheral LPS challenge 1, suggesting these microglia are functionally primed. Thus, microglia priming can refer to either the enhanced responsiveness to a second stimulus (functional definition) or the pre-activated state of microglia compared to naïve microglia.
To be consistent with our previous publication 3, in Chapter 3, we still used the term microglia priming to describe the pre-activated state induced by the intrinsic Ercc1-deficiency. When we compared the transcriptional profile of microglia in
Cx3cr1creERT2:Ercc1ko/loxP mice at 2 months after tamoxifen treatment with the common
7 this result, we concluded that no microglia priming occurs in Cx3cr1creERT2:Ercc1ko/loxP
mice at 2 months after tamoxifen treatment, when a large proportion of cells were still
Ercc1-deficient. However, we observed microglia priming in constitutive Ercc1Δ/ko mice
1. This is possibly due to the fact that the intrinsic DNA damage induced in microglia is limited compared to the global DNA damage accumulation observed in response to overall Ercc1 deletion in all cell types in the brain. Notably, Camk2creER
-driven Ercc1-deletion in forebrain neurons also resulted in a microglia phenotype reminiscent of what was observed in Ercc1Δ/ko mice 14. These results suggest that the neuronal genotoxic stress is a main contributor to microglia priming in constitutive Ercc1Δ/ko
mouse, whereas the intrinsic DNA damage does not induce microglia priming. It is worth noting that, in Chapter 3, if we used the functional definition of microglia priming, the conclusion could be different. Because we did observe enhanced microglia responsiveness to a peripheral LPS challenge in terms of Ccl2 expression in
Cx3cr1creERT2:Ercc1ko/loxP mice at 2 and 6 months after tamoxifen treatment. But
compared to Ercc1Δ/ko mouse microglia, Ercc1-deficient microglia did not show
enhanced responsiveness of Tnf expression to peripheral LPS challenge in
Cx3cr1creERT2:Ercc1ko/loxP mice. These results indicate that intrinsic DNA damage has a
limited influence on microglia priming.
In Chapter 4, we adopted the new terms as defined by Neher and Cunningham to describe microglia innate immune memory. Previous studies have shown that pathogen type, dose, and exposure times could induce different innate immune memories in microglia 4,5,11. In Chapter 4, we further showed that the interval time between two challenges is also crucial for developing microglia innate immune memory. Systemic administration of β-glucan induced immune training in microglia at 2 days after the injection. But, at 7 days and 14 days after the first β-glucan injection, we did not observe immune training in microglia. In addition, LPS induced immune training in microglia at 2 days after the injection. However, 7 days later, we observed immune tolerance rather than training. Transcriptional and epigenetic profiling of microglia at different time points after β-glucan or LPS injection could provide more insight into the underlying mechanisms of this phenomenon.
198
Future perspectives
1) Microglia subtypes in the innate immune response. In Chapter 3 and 4, we used the whole population of microglia from the brain to study microglia innate immune memory. Given the considerable microglia heterogeneity as discussed in Chapter 1, it is likely that the innate immune response is also heterogeneous. This assumption is supported by the emergence of several distinct microglia morphological clusters in LPS/β-glucan preconditioned mice cortex in response to a second LPS challenge in Chapter 4. Additionally, midbrain microglia present a high immune-alert state under steady state but elicit an immune-suppressive response when they are exposed to a systemic LPS injection compared to cortex, hippocampus, and striatum 15. Thus, it would be interesting to investigate the microglia phenotype diversity in innate immune response across different brain regions or within specific brain regions at single-cell resolution.
2) Are morphologically changed microglia immune-trained? In Chapter 4, a microglia morphological cluster related to LPS/β-glucan preconditioning was identified in response to a second LPS stimulus. Nevertheless, whether these morphologically changed microglia are immune-trained is unclear. To address this question, in situ RNA hybridization combined with Iba1 immunohistochemistry or microglia fate mapping systems are needed. The probes can be designed based on the current knowledge of genes whose expression show an enhanced response to a second LPS stimulus in LPS/β-glucan trained microglia (hereafter referred to as trained-gene). Nevertheless, further studies are needed to obtain a more comprehensive trained-gene dataset through transcriptional and epigenetic profiling. Using different probes designed based on the trained-gene dataset, will allow us to determine whether morphologically changed microglia are immune trained or not.
7 The crosstalk between the periphery and the brain in microglia innate immune memory
It is challenging to elucidate the effects of systemic stimuli on microglia innate immune memory in vivo. One reason is that the first systemic stimulus could affect microglia directly by passing the BBB and/or indirectly through inducing cytokines in the periphery. The direct and the indirect exposure might affect microglia innate immune memory differently. Thus, the influence of the peripheral stimulus on the BBB integrity needs to be addressed first. In Chapter 4, the immune training phenotypes induced by β-glucan and LPS were different, indicated by similar but not identical cytokine gene expression responses to a second LPS challenge. In addition, LPS preconditioned mouse microglia showed more pronounced morphological changes than β-glucan preconditioned mouse microglia upon a second LPS challenge. Although there is some controversy, most studies have confirmed that the 1 mg/kg LPS we used in Chapter 4 for i.p. injection could compromise the integrity of the BBB 16,17. For β-glucan, no study has been performed so far to investigate whether β-glucan could cross or compromise the BBB. The direct exposure of microglia to LPS could be one of the reasons why LPS and β-glucan trained microglia in different ways. However, when using 0.5 mg/kg LPS which does not compromise the BBB, Wendeln et al. still observed that an LPS challenge given 1 day before induced microglia training 5. This suggests that microglia training in vivo is not dependent on the direct exposure of microglia to LPS. Most likely, the different cytokine profiles induced by β-glucan and LPS in the periphery explained why they trained microglia differently. To address this question, in Chapter 4, we also determined the acute peripheral response at 3 h after LPS and β-glucan injection via cytokine array. Indeed, different peripheral cytokine profiles were induced by LPS and β-glucan in the serum.
The fact that the first stimulus could also affect how the periphery responds to the second challenge further complicates the study of the effect of a systemic stimulus on microglia innate immune memory in vivo. This will complicate the interpretation of the results as it is uncertain whether the microglia immune response observed is caused by the intrinsic changes induced by the first stimulus or an altered peripheral response after the second stimulus. This could be avoided by giving the second stimulus
200
intracerebrally. Nevertheless, the latter would not mimic the situation where microglia respond to systemic inflammation or infection. In Chapter 4, an interesting phenomenon was observed: at 2 days after the first LPS challenge, microglia showed an immune training response to a second LPS challenge, whereas most cytokines in the periphery showed a tolerance response. With different interval times (1 d or 14 h), other studies also observed this different response between brain and periphery to a second LPS challenge in LPS preconditioned mice based on the production of several cytokines such as TNF, IFN-γ, IL-1β, IL-6, IL-12 5,18,19. These results indicate that past inflammatory experience affects the brain and the periphery differently. However, the current observation of training or tolerance in the periphery is only based on the responses of a limited set of cytokines. In the future, peripheral cytokine profiles need to be further characterized by multiplex cytokine assays.
Future perspectives
1) The microglia biphasic dose-response to β-glucan in vitro. It has been confirmed that primary microglia show a biphasic dose-response to LPS 11. Preconditioning with an extremely low dose of LPS (10-3 pg/ml) induces immune training, whereas preconditioning with a much higher dose (> 10 pg/ml) of LPS induces immune tolerance 11. For β-glucan (Candida albicans-derived), this dose-dependent biphasic response is not very consistent 11. Microglia training was induced by a medium dose (10-1 pg/mL) of β-glucan in a previous study, not by a low dose (10-3 pg/ml) of β-glucan 11. Using a different source of β-glucan (Saccharomyces cerevisiae-derived), we only observed immune tolerance with higher doses of β-glucan (> 105 pg/ml) preconditioning in the BV-2 cell line, but training was not observed at the lowest dose (10-2 pg/ml, unpublished data of our group from Zhang et al.). Thus, the microglia dose- dependent response to β-glucan in vivo still needs further investigation. Different β-glucan sources and microglia types (primary microglia or microglia cell line) should be investigated in the dose-response experiment. 2) Can β-glucan cross or compromise the BBB? The in vitro experiment mimics the direct exposure of microglia to β-glucan. However, so far, it is still unclear whether β-glucan can cross the BBB to gain access to parenchymal microglia. This could for instance be studied by radiolabeling β-glucan. To determine whether β-glucan results in leaky BBB,
7 immunohistochemical detection in brain sections of proteins in plasma and commonly used dyes can be used.
3) The contribution of peripheral cytokine(s) to microglia innate immune memory. Recently, Wendeln et al. showed that one individual cytokine, TNF-alpha, applied peripherally also could induce immune memory in the brain 5. In the future, cytokine profiles need to be further characterized by multiplex cytokine assay after a first challenge. Then, the potential cytokine or chemokine induced by the first challenge could be applied individually or blocked by neutralizing antibodies in the periphery, to study the influence of individual cytokines or chemokines on microglia innate immune memory. And conversely, it would also be a useful experiment to employ mice lacking
202
Biological relevance of innate immune memory
A vaccine is classically designed to confer host protection against a target pathogen. However, vaccines also affect the host response to unrelated pathogens, which is referred to as heterologous effects of vaccination 20 . For example, Bacillus Calmette-Guérin (BCG) is a vaccine primarily used against tuberculosis. But BCG also has beneficial heterologous effects against all-cause mortality 21. Recently, innate immune memory has been shown to play a role in heterologous effects of BCG. BCG induces immune training in NK cells and peripheral blood monocytes, protecting mice from a lethal dose of C. albicans in a T- and B-lymphocyte-independent manner 22,23. In addition, in humans, BCG induces immune training in monocytes, protecting against non-related viral infection of yellow fever vaccine 24. Of note, BCG induced training in monocytes can last at least 3 months after vaccination 22, which is much longer than the lifespan of the monocytes 25. This observation suggests that the reprogramming also takes place in progenitor cells. Indeed, BCG can reprogram hematopoietic stem cells (HSC) and progenitor cells in the bone marrow, enhance their expansion and promote myelopoiesis 26. Importantly, the monocytes and macrophages derived from the bone marrow are also immune trained, protecting mice against Mycobacterium tuberculosis infection 26. It is worth noting that in this study, BCG was given intravenously. When BCG was vaccinated subcutaneously in the mice, the training effects in bone marrow derived macrophages were not observed 26. Similar to this study, Mitroulis et al. showed that i.p. injection of β-glucan could also reprogram HSCs and progenitor cells in the BM, and promote myelopoiesis, protecting mice against a secondary systemic inflammation and chemotherapy 27. Thus, the peripheral stimulus not only affects mature myeloid cells, but also affects the precursors of innate immune cells, providing long term innate immune memory. Besides BCG and β-glucan, preconditioning with other pathogen components was also shown to protect mice against lethal infections or injuries. For example, protective effects against Streptococcus pneumoniae and rotavirus are induced by flagellin (TLR5) preconditioning 20, CpG dinucleotides (TLR9) preconditioning protects against E. coli meningitis 20, and LPS (TLR4) preconditioning against E. coli 28 as well as stroke induced brain injury 29,30. Nevertheless, the preconditioning also sometimes induces detrimental effects. Mice preconditioned with
7 a super-low dose (5 ng/kg body weight) of LPS displayed exacerbated sepsis-induced tissue damage, bacterial load in circulation, and mortality 31. In summary, these observations suggest that the contributions of innate immune memory must be taken into consideration to improve vaccine efficacy and safety 20.
Future perspectives
Modulating innate immune memory in microglia in CNS diseases. As long-lived tissue macrophages in the CNS, microglia innate immune memory induced by a previous stimulus might have long term functional impacts. We previously showed that LPS-induced immune tolerance in microglia could at least last for 32 weeks 4. Wendeln et al. showed that microglia are transcriptionally and epigenetically altered even 6 months after LPS challenges. In addition, previous peripheral LPS challenges could influence Aβ plaque load 6 months later in the brains of APP23 mice 5. A single i.p. LPS injection (0.5 mg/kg) induced immune training in the brain, which exacerbated cerebral β-amyloidosis. In contrast, four LPS (0.5 mg/kg) injections on 4 consecutive days induced immune tolerance in the brain, which alleviated the pathology 5. The beneficial effect of immune tolerance was also observed in a stroke mouse model 5. Recently, in primary microglia, chronic exposure (twice 24 h exposure with an interval of 3-5 days) to Aβ induced immune tolerance and metabolic defects. These defects were characterized by reduced glycolysis and oxidative phosphorylation compared to microglia exposed to Aβ only once 32. In addition, metabolic boosting with IFN-γ could restore immunological function of microglia 32. Similarly, in monocytes, β-glucan partially reversed LPS-induced immune tolerance through epigenetic regulation 33. In Chapter 4, we show that β-glucan could induce immune training in microglia. It would be interesting to investigate whether β-glucan or other immune modifiers could reverse the immune state of microglia in disease conditions. Thus, modulating innate immune memory in
204
Microglia replacement as a treatment for microglia-associated diseases
As discussed in Chapter 1, microglia depletion and repopulation were shown to be a promising strategy for the treatment of neurological disorders. Recently, three new strategies were developed to efficiently replace microglia at a CNS-wide scale or in a specific brain region in mouse 34. The first two strategies are microglia replacement by bone marrow transplantation (mrBMT) and microglia replacement by peripheral blood (mrPB). Both strategies are based on traditional bone marrow transplantation 35 and peripheral blood chimerism models described previously 36, but microglia were depleted prior to the transplantation and irradiation. Strikingly, mrBMT and mrPB can replace 92.66% and 80.74% of the resident microglia in the brain, respectively. mrPB and especially mrBMT cells are largely derived from peripheral CCR2-positive cells 34. However, without previous microglia depletion, the periphery-derived cells were barely detected in the brain of traditional models 35-37, suggesting that an empty microglia niche is required for efficient microglia transplantation. Nevertheless, mrBMT and mrPB cells are different from the resident microglia, indicated by a less ramified morphology and more macrophage-like transcriptional profiles 34. The last strategy is microglia replacement by microglia transplantation (mrMT), which applies traditional microglia transplantation into a microglia-depleted brain. With this strategy, they efficiently engrafted microglia in the brain, and the mrMT cells could be found in the ipsilateral and neighboring regions of the injected site one month after transplantation 34. Importantly, mrMT cells are transcriptionally and morphologically similar to resident microglia 34.
Microglia depletion or replacement can improve neurophysiological and behavioral abnormalities in various disease models 38-42. Most of these microglia replacement studies used CSF1R inhibitors to deplete microglia, as genetic depletion strategies are not clinically feasible. In addition. pharmacologic depletion of microglia by CSF1R inhibitors did not induce behavioral abnormalities in mouse as discussed in Chapter 1. Moreover, CSF1R inhibition was shown to be safe in several phase I studies in cancer patients 43-46. However, in a recent study, microglia were shown to inhibit neuronal activity 47. Microglia depletion by inhibition of the CSF1R amplified and synchronized the activity of neurons, which significantly increased drug-induced seizure responses
7 in mice 47. Thus, before using these microglia replacement strategies to treat neurological disorders related to microglia in human, more extensive investigation of potential adverse effects of microglia depletion in CNS-diseased patients is urgently needed 34.
In Chapter 3, we showed that the Cx3cr1CreERT2-driven Ercc1-deletion in microglia also
resulted in a gradual microglia replacement. Ercc1-deletion led to a progressive loss of microglia cells in the brain from 1 to 6 months after tamoxifen treatment. At 6 months after tamoxifen treatment, around 50% microglia were lost. This gradual loss of microglia is quite different from the fast depletion observed in other genetic depletion systems (Table 2, Chapter 1). This is probably due to the fact that microglia require a certain time span to accumulate DNA damage. In response to the microglia loss, remaining microglia showed increased proliferation, which is likely due to an empty microglia niche as indicated by a previous study 34. From 6 months to 12 months after tamoxifen, microglia maintained 40-50% of their population. However, during this time, more Ercc1-deficient microglia than non-deficient microglia were lost. In the end, non-deficient microglia replaced the microglia pool at 12 months after tamoxifen treatment. Compared to other genetic depletion systems such as CX3CR1CreER: R26iDTR
and CD11b-HSVTK (Table 2, Chapter 1), no overt signs of microgliosis were observed during the repopulation.
Future perspectives
1) The origin of the repopulated microglia in the Cx3cr1creERT2:Ercc1ko/loxP system.
Microglia renew themselves exclusively by self-renewal under physiological conditions, but this is possibly changed by the empty microglia niche 34. Thus, the contribution of the periphery needs to be addressed in the Cx3cr1creERT2:Ercc1ko/loxP mice. In our model,
at 12 months after tamoxifen, microglia from Cx3cr1creERT2:Ercc1ko/loxP mice are
transcriptionally similar to the controls again. However, since the monocyte-derived macrophages in the brain have distinct transcriptional profiles compared to the resident microglia 34, this indicates that the repopulated microglia are unlikely to be derived from the periphery. To further support this conclusion, the
206
no RFP+ microglia are observed during repopulation, it can be excluded that repopulated microglia originate from peripheral CCR2-positive monocytes.
2) The effect of Ercc1-deletion in microglia on other cell types in the brain and animal behavior. In this study, our main focus was to determine the effect of intrinsic DNA damage on microglia. The effect of Ercc1-deletion in microglia on other cell types is still unknown. In addition, behavioral studies in Cx3cr1creERT2:Ercc1ko/loxP mice after
tamoxifen treatment are needed.
3) The effect of microglia replacement by the Cx3cr1creERT2:Ercc1ko/loxP system on
neurophysiological and behavioral abnormalities in disease models. The deletion of Ercc1 first induced a transient ageing phenotype in microglia at 1, 2 and 6 months after tamoxifen treatment. With the progressive replacement of Ercc1-deficient microglia, at 12 months after tamoxifen treatment, the repopulated microglia (95% are Ercc1 non-deficient) were transcriptionally similar to WT microglia again. In addition, we showed that the repopulated microglia at 12 months after tamoxifen had similar proliferation ability and showed a highly similar response to a peripheral LPS challenge as the WT microglia. Thus, the replacement of microglia requires at least 12 months in the
Cx3cr1creERT2:Ercc1ko/loxP system. The long replacement time required compared to other
systems (Table 2, Chapter 1) could be the main factor limiting its use as a microglia replacement system. Moreover, the repopulated Ercc1 non-deficient microglia at 12 months after tamoxifen treatment still showed an enlarged morphology similar to the Ercc1-deficient microglia at 1, 2 or 6 months after tamoxifen treatment. This is probably a compensation effect for the reduced microglia numbers at 12 months after tamoxifen treatment. Besides what we investigated in Chapter 3, it is still unknown whether these enlarged, repopulated, Ercc1 non-deficient microglia at 12 months after tamoxifen treatment are functionally similar to WT microglia in other analyses. This could be another reason why this model is not appropriate for studying microglia replacement
7 Transcriptional profiling of microglia: current state of the art and future perspectives
In the past decade, substantial knowledge of microglia has been gained by transcriptional profiling of these cells. By comparing microglia to other tissue macrophages or monocytes through bulk-population RNA-seq, we now have a better understanding of microglia identity (described in Chapter 1). This is a valuable resource to develop microglia fate mapping tools (e.g. Tmem119tdTomato 48, HexbtdTomato 49, HexbcreERT2 49 and P2ry12creER 50 mouse lines) and microglia-specific antibodies (e.g.
Tmem119 51, P2ry12 52, FCRLS 52 and Siglec-H 53). Through single-cell profiling, CAMs can be distinguished from microglia as described in Chapter 1. In addition, single-cell profiling reveals previously unrecognized phenotypic diversity of microglia both under healthy and disease conditions as described in Chapter 1. Several microglia subpopulations related to disease progression or injury have been identified such as DAM 54, MGnD 55, and ARM 56 in the AD mouse model, disease-associated microglia in the EAE mouse model 57, microglia subtypes related to mouse demyelinating pathologies (cuprizone) and facial nerve axotomy 58, and MS-associated microglia in humans 58
Recently, it has been shown that single-nucleus RNA-seq can faithfully recapitulate transcriptional changes at the tissue level and the single-cell level 8,9,59,60. However, for microglia, it was still unknown whether the nuclear transcriptome can serve as an alternative for the cellular transcriptome, whether the nuclear transcriptome contains enough information to identify microglia subtypes, and whether single-nucleus RNA-seq is amenable to frozen CNS tissues. The data presented in Chapter 5 showed that microglia nuclear transcriptomes generated from both fresh and frozen tissues are a good proxy for cellular transcriptomes, ensuring the use of single-nucleus RNA-seq to study microglia from banked CNS tissues. Of note, the freshly isolated microglia might not fully represent the original donor microglia on the transcriptional level, as microglia might lose their processes through the mechanical dissociation and the cell sorting. Nevertheless, current freshly isolated microglia transcriptomes are already the closest resemblance of human microglia transcriptomes, since the microglia-specific translating ribosome affinity purification (TRAP) approach 61 is not amenable to
208 humans. In Chapter 6, we demonstrated that PMD has very limited influence on mouse and human microglia transcriptomes. We identified 50 PMD-related genes in the mouse, where ante-mortem and postmortem conditions are tightly controlled. Compared to mouse, directly analyzing the PMD effects on human microglia is challenging, in view of the limited sample size (37 microglia samples) and not well documented pre-mortem and postmortem variables in our dataset compared to previous studies 62,63. Thus, we decided to extrapolate the knowledge from mouse to human, investing human homologs of mouse microglia PMD-related genes. We observed that PMD up to 12 h has limited effects on human microglia transcriptomes. However, it is worth noting that PMD in humans might affect more genes than we identified, since we only investigated human homologs of mouse microglia PMD-related genes. Gene expression profiling of isolated cells, either in bulk or at the single cell level, is inherently associated with loss of spatial, contextual information of the used tissue. Relatively recently, two technologies were developed where gene expression profiles were generated while retaining spatial information of the analyzed tissue section. The first technology, spatial transcriptomics 64, makes use of glass slides on which oligo d(T) primers with positional barcodes are spotted to capture the mRNA present in the overlaid tissue. These positional barcodes enable the maintenance of positional information throughout the process of cDNA synthesis, library preparation and sequencing. With decreasing spot diameters and spot distance, resolution will further increase, which is required for single-cell analysis. A second approach to sequencing RNAs in the context of cells and tissues is fluorescent in situ sequencing (FISSEQ) 65. FISSEQ is a technology combining RNA-FISH and next generation sequencing, allowing for the detection of multiple RNAs at subcellular resolution. Where FISSEQ provides a higher resolution than spatial transcriptomics, it requires a panel of oligonucleotides to detect mRNAs of interest where spatial transcriptomics is unbiased and does not require a priori knowledge about genes of interest. Application of these techniques to microglia profiling will provide more insight into microglia transcriptomes in the future.
7 Conclusions
Collectively, the present findings demonstrate that the DNA repair gene Ercc1 is essential for microglia, and its deficiency in microglia induces a transient aging signature, which is different from a priming or disease-associated microglia gene expression profile. The past experience of peripheral inflammatory challenge such as LPS and β-glucan can affect how microglia respond to a sequential stimulus. Thus, microglia responses could be modulated by some immune modifiers, which could provide promising targets to develop treatment for microglia-associated diseases. In addition, the results suggest that both in humans and mice, microglia nuclei are a good proxy to freshly isolated microglia cells on transcriptional level, and postmortem delay has limited influence on microglia transcriptomes. These findings justify the use of banked postmortem human brain specimens to study microglia in neurodegenerative
210
References
1 Raj, D. D. A. et al. Priming of microglia in a DNA-repair deficient model of accelerated aging.
Neurobiol Aging 35, 2147-2160 (2014).
2 Cunningham, C. et al. Central and systemic endotoxin challenges exacerbate the local inflammatory response and increase neuronal death during chronic neurodegeneration. J Neurosci 25, 9275-9284 (2005).
3 Holtman, I. R. et al. Induction of a common microglia gene expression signature by aging and neurodegenerative conditions: a co-expression meta-analysis. Acta Neuropathol Commun 3, 31 (2015).
4 Schaafsma, W. et al. Long-lasting pro-inflammatory suppression of microglia by LPS- preconditioning is mediated by RelB-dependent epigenetic silencing. Brain Behav Immun 48, 205-221 (2015). 5 Wendeln, A. C. et al. Innate immune memory in the brain shapes neurological disease hallmarks. Nature 556, 332-338 (2018). 6 Geirsdottir, L. et al. Cross-species single-cell analysis reveals divergence of the primate microglia program. Cell 179, 1609-1622 (2019). 7 Gosselin, D. et al. An environment-dependent transcriptional network specifies human microglia identity. Science 356 (2017).
8 Hu, P. et al. Dissecting cell-type composition and activity-dependent transcriptional state in mammalian brains by massively parallel single-nucleus RNA-Seq. Mol Cell 68, 1006-1015 (2017). 9 Lake, B. B. et al. Neuronal subtypes and diversity revealed by single-nucleus RNA sequencing of the
human brain. Science 352, 1586-1590 (2016).
10 Galatro, T. F. et al. Transcriptomic analysis of purified human cortical microglia reveals age-associated changes. Nat Neurosci 20, 1162-1171 (2017).
11 Lajqi, T. et al. Memory-like inflammatory responses of microglia to rising doses of LPS: Key role of PI3Kgamma. Front Immunol 10, 2492 (2019).
12 Neher, J. J. & Cunningham, C. Priming microglia for innate immune memory in the brain. Trends
Immunol 40, 358-374 (2019).
13 Haley, M. J. et al. Microglial priming as trained immunity in the brain. Neuroscience 405, 47-54 (2019). 14 Raj, D. D. et al. Priming of microglia in a DNA-repair deficient model of accelerated aging. Neurobiol Aging 35, 2147-2160 (2014). 15 Abellanas, M. A. et al. Midbrain microglia mediate a specific immunosuppressive response under inflammatory conditions. J Neuroinflammation 16, 233 (2019). 16 Varatharaj, A. & Galea, I. The blood-brain barrier in systemic inflammation. Brain Behav Immun 60, 1-12 (2017). 17 Hoogland, I. C. et al. Systemic inflammation and microglial activation: systematic review of animal experiments. J Neuroinflammation 12, 114 (2015). 18 Puntener, U. et al. Long-term impact of systemic bacterial infection on the cerebral vasculature and microglia. J Neuroinflammation 9, 146 (2012).
19 Chen, R. et al. Differential expression of cytokines in the brain and serum during endotoxin tolerance. J Neuroimmunol 163, 53-72 (2005). 20 Goodridge, H. S. et al. Harnessing the beneficial heterologous effects of vaccination. Nature reviews. Immunology 16, 392-400 (2016). 21 Uthayakumar, D. et al. Non-specific effects of vaccines illustrated through the BCG example: From observations to demonstrations. Front Immunol 9, 2869 (2018). 22 Kleinnijenhuis, J. et al. Bacille Calmette-Guerin induces NOD2-dependent nonspecific protection from reinfection via epigenetic reprogramming of monocytes. PNAS 109, 17537-17542 (2012). 23 Kleinnijenhuis, J. et al. BCG-induced trained immunity in NK cells: Role for non-specific protection to infection. Clin Immunol 155, 213-219 (2014). 24 Arts, R. J. W. et al. BCG vaccination protects against experimental viral infection in humans through the induction of cytokines associated with trained immunity. Cell Host Microbe 23, 89-100 (2018). 25 Patel, A. A. et al. The fate and lifespan of human monocyte subsets in steady state and systemic inflammation. J Exp Med 214, 1913-1923 (2017). 26 Kaufmann, E. et al. BCG educates hematopoietic stem cells to generate protective innate immunity against tuberculosis. Cell 172, 176-190 (2018).
7 27 Mitroulis, I. et al. Modulation of myelopoiesis progenitors is an integral component of trained
immunity. Cell 172, 147-161 (2018).
28 Kopanakis, K. et al. Pre-treatment with low-dose endotoxin prolongs survival from experimental lethal endotoxic shock: Benefit for lethal peritonitis by Escherichia coli. Cytokine 62, 382-388 (2013).
29 Rosenzweig, H. L. et al. Endotoxin preconditioning protects against the cytotoxic effects of TNFalpha after stroke: a novel role for TNFalpha in LPS-ischemic tolerance. J Cereb Blood Flow
Metab 27, 1663-1674 (2007).
30 Garcia-Bonilla, L. et al. Endogenous protection from ischemic brain injury by preconditioned monocytes. J Neurosci 38, 6722-6736 (2018).
31 Chen, K. et al. Super-low dose endotoxin pre-conditioning exacerbates sepsis mortality.
EBioMedicine 2, 324-333 (2015).
32 Baik, S. H. et al. A breakdown in metabolic reprogramming causes microglia dysfunction in Alzheimer's Disease. Cell Metab 30, 493-507 (2019).
33 Novakovic, B. et al. Beta-glucan reverses the epigenetic state of LPS-induced immunological tolerance. Cell 167, 1354-1368 (2016). 34 Xu, Z. et al. Efficient strategies for microglia replacement in the central nervous system. Cell Rep 32, 108041 (2020). 35 Ajami, B. et al. Local self-renewal can sustain CNS microglia maintenance and function throughout adult life. Nat Neurosci 10, 1538-1543 (2007). 36 Ajami, B. et al. Infiltrating monocytes trigger EAE progression, but do not contribute to the resident microglia pool. Nat Neurosci 14, 1142-1149 (2011). 37 Askew, K. et al. Coupled proliferation and apoptosis maintain the rapid turnover of microglia in the adult brain. Cell Rep 18, 391-405 (2017).
38 Rice, R. A. et al. Elimination of microglia improves functional outcomes following extensive neuronal loss in the hippocampus. J Neurosci 35, 9977-9989 (2015).
39 Rice, R. A. et al. Microglial repopulation resolves inflammation and promotes brain recovery after injury. Glia 65, 931-944 (2017).
40 Dagher, N. N. et al. Colony-stimulating factor 1 receptor inhibition prevents microglial plaque association and improves cognition in 3xTg-AD mice. J Neuroinflammation 12, 139 (2015). 41
Ikezu, S. et al. Inhibition of colony stimulating factor 1 receptor corrects maternal inflammation-induced microglial and synaptic dysfunction and behavioral abnormalities. Mol Psychiatry doi: 10.1038 (2020). 42 Willis, E. F. et al. Repopulating microglia promote brain repair in an IL-6-dependent manner. Cell 180, 833-846 (2020). 43 Cannarile, M. A. et al. Colony-stimulating factor 1 receptor (CSF1R) inhibitors in cancer therapy. J Immunother Cancer 5, 53 (2017). 44 Lee, J. H. et al. A phase I study of pexidartinib, a colony-stimulating factor 1 receptor inhibitor, in Asian patients with advanced solid tumors. Invest New Drugs 38, 99-110 (2020). 45 Papadopoulos, K. P. et al. First-in-human study of AMG 820, a nonoclonal anti-colony-stimulating factor 1 receptor antibody, in patients with advanced solid tumors. Clin Cancer Res 23, 5703-5710 (2017). 46 Tap, W. D. et al. Structure-guided blockade of CSF1R kinase in tenosynovial giant-cell tumor. N Engl J Med 373, 428-437 (2015). 47 Badimon, A. et al. Negative feedback control of neuronal activity by microglia. Nature 586, 417-423 (2020). 48 Ruan, C. et al. A novel Tmem119-tdTomato reporter mouse model for studying microglia in the central nervous system. Brain Behav Immun 83, 180-191 (2020). 49 Masuda, T. et al. Novel Hexb-based tools for studying microglia in the CNS. Nat Immunol 21, 802-815 (2020). 50 McKinsey, G. L. et al. A new genetic strategy for targeting microglia in development and disease. Elife 9, e54590 (2020). 51 Bennett, M. L. et al. New tools for studying microglia in the mouse and human CNS. PNAS USA 113, E1738-1746 (2016). 52 Butovsky, O. et al. Identification of a unique TGF-β–dependent molecular and functional signature in microglia. Nat Neurosci 17, 131-143 (2014).
53 Konishi, H. et al. Siglec-H is a microglia-specific marker that discriminates microglia from CNS-associated macrophages and CNS-infiltrating monocytes. Glia 65, 1927-1943 (2017).
212
54 Keren-Shaul, H. et al. A unique microglia type associated with restricting development of Alzheimer's Disease. Cell 169, 1276-1290 (2017).
55 Krasemann, S. et al. The TREM2-APOE pathway drives the transcriptional phenotype of dysfunctional microglia in neurodegenerative diseases. Immunity 47, 566-581 (2017). 56 Sala Frigerio, C. et al. The major risk factors for Alzheimer's Disease: Age, sex, and genes modulate the microglia response to Abeta plaques. Cell Rep 27, 1293-1306 (2019). 57 Jordao, M. J. C. et al. Single-cell profiling identifies myeloid cell subsets with distinct fates during neuroinflammation. Science 363, eaat7554 (2019). 58 Masuda, T. et al. Spatial and temporal heterogeneity of mouse and human microglia at single-cell resolution. Nature 566, 388-392 (2019). 59 Habib, N. et al. Div-Seq: Single-nucleus RNA-Seq reveals dynamics of rare adult newborn neurons. Science 353, 925-928 (2016). 60 Lake, B. B. et al. A comparative strategy for single-nucleus and single-cell transcriptomes confirms accuracy in predicted cell-type expression from nuclear RNA. Sci Rep 7, 6031 (2017).
61 Ayata, P. et al. Epigenetic regulation of brain region-specific microglia clearance activity. Nat
Neurosci 21, 1049-1060 (2018).
62 Ferreira, P. G. et al. The effects of death and postmortem cold ischemia on human tissue transcriptomes. Nat Commun 9, 490 (2018).
63 Zhu, Y. et al. Systematic analysis of gene expression patterns associated with postmortem interval in human tissues. Sci Rep 7, 5435 (2017).
64 Stahl, P. L. et al. Visualization and analysis of gene expression in tissue sections by spatial transcriptomics. Science 353, 78-82 (2016).
65 Lee, J. H. et al. Highly multiplexed subcellular RNA sequencing in situ. Science 343, 1360-1363 (2014).
