Search

Sepsis is a severe and frequent disease in modern intensive medicine. Sepsis does not only present with a mortality rate of 30% of all patients, but survivors often show long lasting cognitive deficits, loss of independency and reduced quality of life. However, the mechanisms of sepsis-induced cognitive impairments remain incompletely understood so far. To understand the mechanisms of cognitive decline in sepsis survivors, the possible involvement of the brain’s primary immune cells, microglia, was investigated in this project. In particular, to investigate if and how microglia contribute to synapse loss acutely and long-term after sepsis, different phagocytosis-deficient mice strains were analyzed for synapse number and phagocytosed synapse number. To model sepsis, female mice were injected on two consecutive days with either 1.5 mg/kg body weight lipopolysaccharide (LPS), a bacterial cell wall component that induces strong inflammatory responses. At different timepoints (1 day, 3 days, 7 days or 2 months) tissue was then analysed. Acutely after LPS-injections, quantification of phagocytic proteins showed a sharp increase in complement protein 3 in the brain. Consistently, immunohistochemical analysis revealed, that functional synapse number was strongly reduced and that the percentage of phagocytosed synapses was increased by LPS treatment. Interestingly, mice deficient in phagocytic signaling proteins (Mfge8 -/-, Cd11b-/-) showed no reduction in functional synapse number and no increase in phagocytosis of functional synapses after LPS-injections. Therefore, microglial phagocytosis leads to synapse loss in a mouse model of sepsis and may contribute to long-term cognitive deficits in sepsis survivors. Inhibition of specific microglial pathways could therefore be a promising new strategy to prevent long-term cognitive deficits in sepsis survivors.

Due to the ever-increasing age of the human population, the absolute number of individuals impacted by neurodegenerative diseases has been growing globally for the last 25 years. As most of them are currently incurable, neurological disorders are the primary cause of disability and death worldwide. Thus, a detailed understanding of the underlying processes of neurodegenerative disorders and the search for therapies are becoming increasingly urgent tasks. Parkinson’s disease (PD) and Huntington’s disease (HD) are chronic neurodegenerative conditions with neuronal loss in the motor, sensory, and cognitive systems. Increasing evidence indicates that neuroinflammation plays a substantial role in the pathogenesis of various neurodegenerative disorders. Therefore, it is crucial to understand how exactly the brain’s immune system affects neuropathology and whether environmental factors that change the immune system response could be risk factors for neurodegenerative diseases. Recent research revealed that innate immune memory in microglia could shape cerebral β-amyloidosis in an AD mouse model and can modify brain cytokine levels, microglial epigenetic and gene expression profiles, metabolism, and phagocytic activity for at least 6 months. However, whether microglial immune memory influences disease pathogenesis in models of PD and HD remains unknown. Therefore, the aim of this thesis was to investigate the effect of innate immune memory on microglial transcriptomic profiles in PD and HD rat models. To this end, two well-characterized rat models of a-synucleinopathy or mutant huntingtin (mHTT) pathology, the neuropathological hallmarks of PD or HD, respectively, were used. The first set of experiments of my thesis focused on investigating whether immune memory can be induced in the blood and brain by peripheral immune stimuli (using one, two, or four injections of bacterial lipopolysaccharides) and a high-fat diet in PD and HD rat models. The second set of experiments of my thesis focused on investigating whether immune memory triggered by peripheral stimuli and a high-fat diet is apparent in the microglial transcriptomic profiles in PD and HD rat models. The third set of experiments of my thesis focused on the heterogeneity of microglial cells in 9- month-old BAC HD and BAC SNCA rats using single-cell RNA-sequencing (scRNA-seq). The findings of my thesis support the existence of innate immune memory in microglia across a variety of disease models and species and provide the first evidence that the same initial immune stimulus causes brain pathology-specific modulation of transcriptomic responses in microglia (as evidenced by different molecular pathways being modulated in PD vs HD models here). The results of the single-cell experiments suggest for the first time that only specific microglial subpopulations react in the early stages of PD and HD pathology; whether these cells are associated with driving or preventing pathology requires further investigation.

Amyloids occur as localized or systemic deposits composed of compact β-sheet fibrils of abnormally folded proteins that may disrupt the physiological function of the afflicted tissue by loss-of-function or gain-of-toxicity. The best characterized amyloid, amyloid-β (Aβ) is a known trigger of the molecular and cellular cascade leading to Alzheimer’s disease (AD). Aβ aggregation can amplify via a prion-like seeding mechanism and rapidly spread in the brain connectome. Several downstream processes potentiate the toxic cascade, ultimately resulting in brain damage and memory loss. Interestingly, vascular dysfunction is described as an early symptom of AD in humans. Recently, a study identified another protein that deposits in the vasculature of the AD brain – Medin amyloid (an internal fragment of MFG-E8). Medin is the most common human vascular amyloid known to date, as it can be detected within the thoracic aorta and the upper body arteries of nearly everyone above 50 years of age. As the name Medin suggests, it predominantly deposits along the elastic fibers of the tunica media. Because of its vascular localization and potential role in arterial aging it has been suggested as a possible risk factor of AD. However, despite its high prevalence in the aging population, no causal role in human pathogenesis was previously reported. Nevertheless, in vitro studies have described common structural, physicochemical and cytotoxic amyloid properties for Medin, and analyses of human tissue derived from autopsy or surgery implicate a role for Medin in vascular diseases. However, the question of a mechanistic link between the presence of Medin, cerebrovascular dysfunction and AD pathology remains elusive due to lack of appropriate experimental in vivo models, which would enable mechanistic studies. In this dissertation, a suitable animal model to investigate the influence of Medin aggregates on cerebrovascular function was established. In particular, we found that with advancing age wild-type mice develop endogenous Medin aggregates in the aorta and cerebral arteries that resemble the sequence, biochemical properties, and morphology of human Medin amyloid. We also showed in our first study that age-associated decline in cerebral vascular function (evident by impaired artery elasticity) in living mice was rescued by prevention of vascular Medin aggregate formation through a genetic knockout of the Medin-containing C2 domain of MFG-E8 (Mfge8 C2 KO). These observations implicate Medin as a causal factor in age-associated cerebrovascular dysfunction for the first time. Given the fact that Medin was recently shown to template aggregation of serum amyloid A, our second study addressed whether potential co-aggregation of Medin and Aβ may exacerbate AD pathology. Indeed, our study provided first evidence for a direct amyloid-amyloid interaction of Medin and Aβ in two mouse models for cerebral β-amyloidosis. Medin did not only co-localize substantially with Aβ deposits in the brain, but the genetic deletion of Medin also altered plaque and fibril structure in vitro and in vivo, prevented MFG-E8/Medin accumulation within plaques and vasculature, and slowed the onset of pathology in cerebral β-amyloidosis. It is further important to note that in line with its vascular localization in humans, genetic Medin deficiency reduced cerebral β-amyloid angiopathy (CAA) and related microhemorrhages even at end-stage amyloid pathology. Additionally, aorta-derived Medin 'seeds' promoted pre-mature Aβ aggregation in the brain, suggesting that Medin promotes Aβ by a heterologous seeding mechanism. To analyze the translational relevance of a potential heterologous seeding mechanism between Medin and Aβ, we also examined data from the dorsolateral prefrontal cortex of >500 patients from the ROS/MAP cohorts, two longitudinal clinical studies on aging, cognitive decline and AD. Importantly, AD patients showed significantly increased MFGE8 expression compared to non-demented patients. Moreover, MFGE8 gene expression levels predicted cognitive decline independent of amyloid neuropathology, namely Aβ plaques and neurofibrillary tangles. Notably, Medin was previously reported to be increased within cerebral arteries of AD patients in comparison to non-demented controls, and levels of arteriole Medin predicted risk for AD independent of amyloid pathology, age, sex and ApoE status. In summary, this thesis demonstrates the importance to develop new technologies and models to study amyloid pathobiology and cross-talk, as amyloids contribute to common age-related damage and dysfunction, but also affect progression, heterogeneity and co-morbidity of other diseases. Our studies demonstrate a new mechanism that may drive age-associated vascular disease and cerebral β-amyloid angiopathy, highlighting Medin as a potential therapeutic target to maintain vascular health and cognitive function with age.

Alzheimer’s disease (AD) is an age-related neurodegenerative disorder and the most common form of dementia. Thereby, the abnormal deposition of the amyloid-β (Aβ) peptide into plaques is considered to be the primary neuropathological insult in AD. For a small proportion of all AD cases it is well known that rare genetic mutations are causative for very early Aβ deposition (familial Alzheimer’s disease). However, the vast majority of all AD cases manifest at later ages (late-onset Alzheimer’s disease (LOAD)) and are most likely caused by an interplay of multiple genetic variants and the environment. During the last ten years, genome-wide association studies revealed several risk loci that increase the susceptibility for LOAD, and interestingly, many of these genetic variants were found to be associated with innate immune functions of which the resident tissue macrophages of the brain – the microglia – are prime regulators. In general, the innate immune response mediated by the resident tissue macrophages is considered protective as it induces the production of inflammatory modulators and enables phagocytosis and killing of pathogens to prevent further tissue damage. However in the AD brain, the progressive accumulation of Aβ deposits leads to a chronic exposure of microglia to Aβ aggregates and induces an excessive neuro-inflammatory response that is thought to promote disease progression. Interestingly, microglia display a highly plastic phenotype and studies from peripheral tissue macrophages reported that a variety of environmental stimuli can determine but also reprogram their functional phenotype. To this end, this thesis summarizes three different approaches, which aimed to understand but also modulate the myeloid cell immune function during AD with regard to their effects on the pathology of cerebral β-amyloidosis. To begin with, we examined whether peripheral monocytes, which were previously shown to adopt a microglia-like phenotype in the healthy brain, can replace dysfunctional microglia in brains of two different mouse models of cerebral β-amyloidosis and may then restrict Aβ accumulation. For this purpose, we depleted microglia in APPPS1 and APP23 transgenic (tg) mice that expressed the herpes simplex virus thymidine kinase (HSVTK) under the myeloid-cell specific CD11b promoter; the application of the thymidine kinase substrate ganciclovir (GCV), which is converted into a cytotoxic product, then induced microglial death. After a two-week ganciclovir treatment, application was discontinued from two weeks up to six months to allow the peripheral monocytes to repopulate the brain. Interestingly, during the first weeks of repopulation the number of infiltrated monocytes were twice the number of resident microglia in control mice, but the engrafted monocytes failed to cluster around Aβ plaques. Consequently, we did not observe alterations in plaque pathology. Also, a pro-longed incubation for up to six months did not change Aβ load. However, long-term monocyte engraftment for five months induced in pre-depositing APP23 mice enabled the infiltrated monocytes to behave most similar to resident microglia: they began clustering around Aβ depositions, the cell number was virtually equal to control mice and plaque-associated monocytes were TREM2-positive. However, these cells also failed to alter Aβ plaque load. This work indicates that the tissue environment in the brain dominates over myeloid cell origin and thus reprograms myeloid cells to match the resident microglia population, however without prevention of Aβ pathology. Recent studies provide evidence that cells of the innate immune system can, similar to the adaptive immune cells, acquire immunological memory. In particular, a distinct set of primary immune stimuli can either enhance or suppress a subsequent immune response, which is referred to as “training” and “tolerance”, respectively. In a second study, we tested the applicability of the immune memory concept to microglia and examined if the induction of innate immune memory can induce long-lasting changes in the brain’s immune response and thereby alter pathology of neurological diseases. To this end, we injected two different doses of the endotoxin lipopolysaccharide (LPS) into pre-depositing APP23 mice. Whereas a single LPS injection was identified to induce acute training effects, consecutive injections for four days induced tolerance effects in microglia. Accordingly, in the brain, we acutely measured initially enhanced concentrations of inflammatory cytokines which decreased with further LPS injections. When we examined the long-lasting effects of the induced immune memory on Aβ pathology and cortical ischemia at the later time points, the initial training stimulus increased while the tolerance stimulus reduced pathology, which was reflected by changes in Aβ plaque load and neuronal damage, respectively. Immune memory in macrophages was previously shown to be mediated by epigenetic changes in enhancer regions that either stimulate or prevent gene transcription. In accordance, we performed chromatin immunoprecipitation sequencing for histone modifications in isolated microglia to determine changes in their enhancer landscape. Notably, we identified the active enhancer repertoire for hypoxia-inducible factor 1α (HIF-1α), a key modulator for macrophage inflammatory responses, to be enriched in microglia after the induction of trained immune memory (1xLPS). In contrast, pathways related to phagocytic functions showed an increase in active enhancers in the 4xLPS treatment group. Importantly, these epigenetic alterations were reflected by expression changes in the respective genes in the isolated microglia population. By this study, we provide first evidence for long-lasting innate immune memory in the brain that can shape neurological disease outcome and is driven by epigenetic modifications of the microglial enhancer landscape. In a last study, we focused on the microglial phagocytic capacity as an important factor for the modulation of Aβ plaque pathology, as in vitro experiments have reported that microglia can bind to, and engulf Aβ fibrils. However, so far, in vivo studies have not convincingly confirmed these results. Therefore, we investigated the role of the soluble milk fat globuleepidermal growth-factor 8 (MFG-E8) protein, that was recently hypothesized to mediate Aβ phagocytosis in AD pathology. To test the in vivo function of MFG-E8, we crossed mice expressing a functional knockout variant of Mfge8 (Mfge8-/-) with the APPPS1 and APP23 tg mouse models of cerebral β-amyloidosis. In contrast to previous reports, our results indicated that the depletion of MFG-E8 has no impact on Aβ uptake by microglia or subsequent Aβ degradation processes. However, contrary to our expectations, MFG-E8 deficiency reduced Aβ plaque load and Aβ levels in both mouse models without affecting amyloid precursor protein (APP) processing. When we immunohistochemically analyzed MFG-E8 distribution in the brain we observed a strong accumulation of MFG-E8 with congophilic Aβ deposits and co-staining of MFG-E8 with Aβ even showed a partial co-localization of both proteins at the sites of Aβ plaques. While the mechanism of these effects requires further studies, our results suggests that a direct interaction between MFG-E8 and Aβ promotes amyloid aggregation. Taken together, these studies examined different ways of modulating the microglial immune response during AD pathology. Interestingly, the replacement of dysfunctional microglia by peripheral monocytes in the diseased brain did not modify Aβ deposition although the infiltrated monocytes adopted features of plaque-associated microglia. However, when we applied the concept of innate immune memory to the brain through the remodeling of the innate immune response by epigenetic reprogramming of the microglial enhancer repertoire, we identified a promising approach to modify Aβ pathology. Especially the induction of a microglial tolerance state had beneficial long-term effects on the pathology of cerebral β- amyloidosis while training aggravated disease outcome. These results provide, for the first time, evidence that long-lasting modulation of the innate immune reaction may occur due to immunological priming – a mechanism that introduces new targets for dampening Aβ pathology in Alzheimer’s disease. However, in contrast, a direct modification of microglial Aβ phagocytosis through the knockout of Mfge8 is most likely not sufficient to modulate microglia function in AD.

Microglia comprise the resident tissue macrophage population in the brain parenchyma. They acquire a variety of functions in health and disease and are thus one of the most studied cell types regarding their contribution to neurological disease. In Alzheimer’s disease, deposition of amyloid-β occurs in senile plaques and leads to alterations of the microglial phenotype. This feature is mimicked in transgenic mouse models developing cerebral β-amyloidosis. How microglia influence onset and progression of Alzheimer’s disease remains incompletely understood despite many studies. Microglia are developmentally distinct from tissue macrophages in other organs and invade the brain early in embryogenesis. A long lifetime has been attributed to microglia; however, this has only scarcely been studied. Today, technological advances in the development of transgenic mouse models and in vivo microscopy allow for the in depth analysis of microglial turnover. In the first study presented here, we used genetic tools to label individual microglia and followed them in vivo by two-photon imaging. Long-term imaging over many months revealed that microglia are a long-lived cell population with low turnover rates. Furthermore, the rates of microglial proliferation and death were equal, matching previous reports that microglial cell numbers stay stable over time. In line with previous studies, higher proliferation rates were found in a mouse model of cerebral β-amyloidosis. Nevertheless, both in health and disease conditions, microglia lived many months and a proportion of microglia persisted throughout the entire lifespan of mice. The long lifetime of microglia renders them suitable for the characterization of innate immune memory, a recently emerged concept, in the brain. While it was previously believed that only adaptive immune cells depict memory characteristics, recent reports have demonstrated immune memory also in cells of the innate immune system. Herein, a priming immune stimulus leads to a long-lasting change in the activation state of the innate immune cell, thereby modifying its response towards a secondary stimulus. In particular, two paradigms can be distinguished: a heightened response has been termed training, while a reduced response is called tolerance. Our second study aimed to reveal whether training and tolerance are inducible in the brain and could be long-lasting modifiers of later occurring neurological disease pathology. We demonstrate that following peripheral immune stimulation with lipopolysaccharides or certain cytokines acute training and tolerance effects were evident in the brain. Furthermore, training and tolerance had opposing effects on much later occurring neuropathology of stroke and early stages of cerebral β-amyloidosis. Microglial enhancer landscape, gene expression, and function differed in response to peripheral immune stimulation with lipopolysaccharides, indicating long-term microglial reprogramming. Therefore, our study provides evidence for long-lasting innate immune memory in microglia that is epigenetically encoded and is sufficient to alter later developing neuropathology. In the third study, we investigated whether training and tolerance modify amyloid plaque structure and neurotoxicity in mice with more advanced cerebral β-amyloidosis. Strikingly, modifications of amyloid plaque structure were evident 9 months after induction of training and tolerance, and occurred concomitantly with alterations in microglial function. In particular, peripheral training and tolerance stimuli altered the phenotype of plaque-associated microglia or reduced their number, respectively. Both impaired microglial barrier function, a recently described feature of microglia around amyloid plaques in Alzheimer’s disease mouse models and patients. Importantly, modulation of microglial function, as induced by peripheral immune stimulation, altered amyloid plaque structure and thereby increased plaque-associated neurotoxicity. These results indicate that changes in microglial function contribute to neuronal damage by modulating amyloid structure and that such changes may occur in response to peripheral inflammatory events. In summary, this thesis demonstrates that microglia are exceptionally long-lived cells with the capability to form persisting immune memory. Peripheral immune stimulation epigenetically reprograms microglia to acquire distinct phenotypes that subsequently modify pathological features of cerebral β-amyloidosis. Therefore, we identified epigenetic microglial memory of peripheral inflammation as an essential modifier of neurological diseases.