The endotoxin hypothesis of Alzheimer's disease

Lipopolysaccharide (LPS) constitutes much of the surface of Gram-negative bacteria, and if LPS enters the human body or brain can induce inflammation and act as an endotoxin. We outline the hypothesis here that LPS may contribute to the pathophysiology of Alzheimer's disease (AD) via peripheral infections or gut dysfunction elevating LPS levels in blood and brain, which promotes: amyloid pathology, tau pathology and microglial activation, contributing to the neurodegeneration of AD. The evidence supporting this hypothesis includes: i) blood and brain levels of LPS are elevated in AD patients, ii) AD risk factors increase LPS levels or response, iii) LPS induces Aβ expression, aggregation, inflammation and neurotoxicity, iv) LPS induces TAU phosphorylation, aggregation and spreading, v) LPS induces microglial priming, activation and neurotoxicity, and vi) blood LPS induces loss of synapses, neurons and memory in AD mouse models, and cognitive dysfunction in humans. However, to test the hypothesis, it is necessary to test whether reducing blood LPS reduces AD risk or progression. If the LPS endotoxin hypothesis is correct, then treatments might include: reducing infections, changing gut microbiome, reducing leaky gut, decreasing blood LPS, or blocking LPS response.

Cell death by phagocytosis

Cells can die as a consequence of being phagocytosed by other cells - a form of cell death that has been called phagotrophy, cell cannibalism, programmed cell removal and primary phagocytosis. However, these are all different manifestations of cell death by phagocytosis (termed 'phagoptosis' for short). The engulfed cells die as a result of cytotoxic oxidants, peptides and degradative enzymes within acidic phagolysosomes. Cell death by phagocytosis was discovered by Metchnikov in the 1880s, but was neglected until recently. It is now known to contribute to developmental cell death in nematodes, Drosophila and mammals, and is central to innate and adaptive immunity against pathogens. Cell death by phagocytosis mediates physiological turnover of erythrocytes and other leucocytes, making it the most abundant form of cell death in the mammalian body. Immunity against cancer is also partly mediated by macrophage phagocytosis of cancer cells, but cancer cells can also phagocytose host cells and other cancer cells in order to survive. Recent evidence indicates neurodegeneration and other neuropathologies can be mediated by microglial phagocytosis of stressed neurons. Thus, despite cell death by phagocytosis being poorly recognized, it is one of the oldest, commonest and most important forms of cell death.

Activated microglia release β-galactosidase that promotes inflammatory neurodegeneration

Beta (β)-galactosidase is a lysosomal enzyme that removes terminal galactose residues from glycolipids and glycoproteins. It is upregulated in, and used as a marker for, senescent cells. Microglia are brain macrophages implicated in neurodegeneration, and can upregulate β-galactosidase when senescent. We find that inflammatory activation of microglia induced by lipopolysaccharide results in translocation of β-galactosidase to the cell surface and release into the medium. Similarly, microglia in aged mouse brains appear to have more β-galactosidase on their surface. Addition of β-galactosidase to neuronal-glial cultures causes microglial activation and neuronal loss mediated by microglia. Inhibition of β-galactosidase in neuronal-glial cultures reduces inflammation and neuronal loss induced by lipopolysaccharide. Thus, activated microglia release β-galactosidase that promotes microglial-mediated neurodegeneration which is prevented by inhibition of β-galactosidase.

P2Y6 Receptor-Dependent Microglial Phagocytosis of Synapses during Development Regulates Synapse Density and Memory

During brain development, excess synapses are pruned (i.e., removed), in part by microglial phagocytosis, and dysregulated synaptic pruning can lead to behavioral deficits. The P2Y6 receptor (P2Y6R) is known to regulate microglial phagocytosis of neurons, and to regulate microglial phagocytosis of synapses in cell culture and in vivo during aging. However, currently it is unknown whether P2Y6R regulates synaptic pruning during development. Here, we show that P2Y6R KO mice of both sexes had strongly reduced microglial internalization of synaptic material, measured as Vglut1 within CD68-staining lysosomes of microglia at postnatal day 30 (P30), suggesting reduced microglial phagocytosis of synapses. Consistent with this, we found an increased density of synapses in the somatosensory cortex and the CA3 region and dentate gyrus of the hippocampus at P30. We also show that adult P2Y6R KO mice have impaired short- and long-term spatial memory and impaired short- and long-term recognition memory compared with WT mice, as measured by novel location recognition, novel object recognition, and Y-maze memory tests. Overall, this indicates that P2Y6R regulates microglial phagocytosis of synapses during development, and this contributes to memory capacity.SIGNIFICANCE STATEMENT The P2Y6 receptor (P2Y6R) is activated by uridine diphosphate released by neurons, inducing microglial phagocytosis of such neurons or synapses. We tested whether P2Y6R regulates developmental synaptic pruning in mice and found that P2Y6R KO mice have reduced synaptic material within microglial lysosomes, and increased synaptic density in the brains of postnatal day 30 mice, consistent with reduced synaptic pruning during development. We also found that adult P2Y6R KO mice had reduced memory, consistent with persistent deficits in brain function, resulting from impaired synaptic pruning. Overall, the results suggest that P2Y6R mediates microglial phagocytosis of synapses during development, and the absence of this results in memory deficits in the adult.

LRPAP1 is released from activated microglia and inhibits microglial phagocytosis and amyloid beta aggregation

Low-density lipoprotein receptor-related protein-associated protein 1 (LRPAP1), also known as receptor associated protein (RAP), is an endoplasmic reticulum (ER) chaperone and inhibitor of LDL receptor related protein 1 (LRP1) and related receptors. These receptors have dozens of physiological ligands and cell functions, but it is not known whether cells release LRPAP1 physiologically at levels that regulate these receptors and cell functions. We used mouse BV-2 and human CHME3 microglial cell lines, and found that microglia released nanomolar levels of LRPAP1 when inflammatory activated by lipopolysaccharide or when ER stressed by tunicamycin. LRPAP1 was found on the surface of live activated and non-activated microglia, and anti-LRPAP1 antibodies induced internalization. Addition of 10 nM LRPAP1 inhibited microglial phagocytosis of isolated synapses and cells, and the uptake of Aβ. LRPAP1 also inhibited Aβ aggregation in vitro. Thus, activated and stressed microglia release LRPAP1 levels that can inhibit phagocytosis, Aβ uptake and Aβ aggregation. We conclude that LRPAP1 release may regulate microglial functions and Aβ pathology, and more generally that extracellular LRPAP1 may be a physiological and pathological regulator of a wide range of cell functions.

Extracellular tau stimulates phagocytosis of living neurons by activated microglia via Toll-like 4 receptor-NLRP3 inflammasome-caspase-1 signalling axis

In tauopathies, abnormal deposition of intracellular tau protein followed by gradual elevation of tau in cerebrospinal fluids and neuronal loss has been documented, however, the mechanism how actually neurons die under tau pathology is largely unknown. We have previously shown that extracellular tau protein (2N4R isoform) can stimulate microglia to phagocytose live neurons, i.e. cause neuronal death by primary phagocytosis, also known as phagoptosis. Here we show that tau protein induced caspase-1 activation in microglial cells via 'Toll-like' 4 (TLR4) receptors and neutral sphingomyelinase. Tau-induced neuronal loss was blocked by caspase-1 inhibitors (Ac-YVAD-CHO and VX-765) as well as by TLR4 antibodies. Inhibition of caspase-1 by Ac-YVAD-CHO prevented tau-induced exposure of phosphatidylserine on the outer leaflet of neuronal membranes and reduced microglial phagocytic activity. We also show that suppression of NLRP3 inflammasome, which is down-stream of TLR4 receptors and mediates caspase-1 activation, by a specific inhibitor (MCC550) also prevented tau-induced neuronal loss. Moreover, NADPH oxidase is also involved in tau-induced neurotoxicity since neuronal loss was abolished by its pharmacological inhibitor. Overall, our data indicate that extracellular tau protein stimulates microglia to phagocytose live neurons via Toll-like 4 receptor-NLRP3 inflammasome-caspase-1 axis and NADPH oxidase, each of which may serve as a potential molecular target for pharmacological treatment of tauopathies.

The Endotoxin Hypothesis of Parkinson's Disease

The endotoxin hypothesis of Parkinson's disease (PD) is the idea that lipopolysaccharide (LPS) endotoxins contribute to the pathogenesis of this disorder. LPS endotoxins are found in, and released from, the outer membrane of Gram-negative bacteria, for example in the gut. It is proposed that gut dysfunction in early PD leads to elevated LPS levels in the gut wall and blood, which promotes both α-synuclein aggregation in the enteric neurons and a peripheral inflammatory response. Communication to the brain via circulating LPS and cytokines in the blood and/or the gut-brain axis leads to neuroinflammation and spreading of α-synuclein pathology, exacerbating neurodegeneration in brainstem nuclei and loss of dopaminergic neurons in the substantia nigra, and manifesting in the clinical symptoms of PD. The evidence supporting this hypothesis includes: (1) gut dysfunction, permeability, and bacterial changes occur early in PD, (2) serum levels of LPS are increased in a proportion of PD patients, (3) LPS induces α-synuclein expression, aggregation, and neurotoxicity, (4) LPS causes activation of peripheral monocytes leading to inflammatory cytokine production, and (5) blood LPS causes brain inflammation and specific loss of midbrain dopaminergic neurons, mediated by microglia. If the hypothesis is correct, then treatment options might include: (1) changing the gut microbiome, (2) reducing gut permeability, (3) reducing circulating LPS levels, or (4) blocking the response of immune cells and microglia to LPS. However, the hypothesis has a number of limitations and requires further testing, in particular whether reducing LPS levels can reduce PD incidence, progression, or severity. © 2023 The Authors. Movement Disorders published by Wiley Periodicals LLC on behalf of International Parkinson and Movement Disorder Society.

Alzheimer's disease-associated R47H TREM2 increases, but wild-type TREM2 decreases, microglial phagocytosis of synaptosomes and neuronal loss

Triggering receptor on myeloid cells 2 (TREM2) is an innate immune receptor, upregulated on the surface of microglia associated with amyloid plaques in Alzheimer's disease (AD). Individuals heterozygous for the R47H variant of TREM2 have greatly increased risk of developing AD. We examined the effects of wild-type (WT), R47H and knock-out (KO) of human TREM2 expression in three microglial cell systems. Addition of mouse BV-2 microglia expressing R47H TREM2 to primary mouse neuronal cultures caused neuronal loss, not observed with WT TREM2. Neuronal loss was prevented by using annexin V to block exposed phosphatidylserine, an eat-me signal and ligand of TREM2, suggesting loss was mediated by microglial phagocytosis of neurons exposing phosphatidylserine. Addition of human CHME-3 microglia expressing R47H TREM2 to LUHMES neuronal-like cells also caused loss compared to WT TREM2. Expression of R47H TREM2 in BV-2 and CHME-3 microglia increased their uptake of phosphatidylserine-beads and synaptosomes versus WT TREM2. Human iPSC-derived microglia with heterozygous R47H TREM2 had increased phagocytosis of synaptosomes vs common-variant TREM2. Additionally, phosphatidylserine liposomes increased activation of human iPSC-derived microglia expressing homozygous R47H TREM2 versus common-variant TREM2. Finally, overexpression of TREM2 in CHME-3 microglia caused increased expression of cystatin F, a cysteine protease inhibitor, and knock-down of cystatin F increased CHME-3 uptake of phosphatidylserine-beads. Together, these data suggest that R47H TREM2 may increase AD risk by increasing phagocytosis of synapses and neurons via greater activation by phosphatidylserine and that WT TREM2 may decrease microglial phagocytosis of synapses and neurons via cystatin F.

Syk inhibitors protect against microglia-mediated neuronal loss in culture

Microglia are brain macrophages and play beneficial and/or detrimental roles in many brain pathologies because of their inflammatory and phagocytic activity. Microglial inflammation and phagocytosis are thought to be regulated by spleen tyrosine kinase (Syk), which is activated by multiple microglial receptors, including TREM2 (Triggering Receptor Expressed on Myeloid Cells 2), implicated in neurodegeneration. Here, we have tested whether Syk inhibitors can prevent microglia-dependent neurodegeneration induced by lipopolysaccharide (LPS) in primary neuron-glia cultures. We found that the Syk inhibitors BAY61-3606 and P505-15 (at 1 and 10 μM, respectively) completely prevented the neuronal loss induced by LPS, which was microglia-dependent. Syk inhibition also prevented the spontaneous loss of neurons from older neuron-glia cultures. In the absence of LPS, Syk inhibition depleted microglia from the cultures and induced some microglial death. However, in the presence of LPS, Syk inhibition had relatively little effect on microglial density (reduced by 0-30%) and opposing effects on the release of two pro-inflammatory cytokines (IL-6 decreased by about 45%, TNFα increased by 80%). Syk inhibition also had no effect on the morphological transition of microglia exposed to LPS. On the other hand, inhibition of Syk reduced microglial phagocytosis of beads, synapses and neurons. Thus, Syk inhibition in this model is most likely neuroprotective by reducing microglial phagocytosis, however, the reduced microglial density and IL-6 release may also contribute. This work adds to increasing evidence that Syk is a key regulator of the microglial contribution to neurodegenerative disease and suggests that Syk inhibitors may be used to prevent excessive microglial phagocytosis of synapses and neurons.

P2Y6 receptor-dependent microglial phagocytosis of synapses mediates synaptic and memory loss in aging

Aging causes loss of brain synapses and memory, and microglial phagocytosis of synapses may contribute to this loss. Stressed neurons can release the nucleotide UTP, which is rapidly converted into UDP, that in turn activates the P2Y₆ receptor (P2Y₆ R) on the surface of microglia, inducing microglial phagocytosis of neurons. However, whether the activation of P2Y₆ R affects microglial phagocytosis of synapses is unknown. We show here that inactivation of P2Y₆ R decreases microglial phagocytosis of isolated synapses (synaptosomes) and synaptic loss in neuronal-glial co-cultures. In vivo, wild-type mice aged from 4 to 17 months exhibited reduced synaptic density in cortical and hippocampal regions, which correlated with increased internalization of synaptic material within microglia. However, this aging-induced synaptic loss and internalization were absent in P2Y₆ R knockout mice, and these mice also lacked any aging-induced memory loss. Thus, P2Y₆ R appears to mediate aging-induced loss of synapses and memory by increasing microglial phagocytosis of synapses. Consequently, blocking P2Y₆ R has the potential to prevent age-associated memory impairment.

Microglia states and nomenclature: A field at its crossroads

Microglial research has advanced considerably in recent decades yet has been constrained by a rolling series of dichotomies such as "resting versus activated" and "M1 versus M2." This dualistic classification of good or bad microglia is inconsistent with the wide repertoire of microglial states and functions in development, plasticity, aging, and diseases that were elucidated in recent years. New designations continuously arising in an attempt to describe the different microglial states, notably defined using transcriptomics and proteomics, may easily lead to a misleading, although unintentional, coupling of categories and functions. To address these issues, we assembled a group of multidisciplinary experts to discuss our current understanding of microglial states as a dynamic concept and the importance of addressing microglial function. Here, we provide a conceptual framework and recommendations on the use of microglial nomenclature for researchers, reviewers, and editors, which will serve as the foundations for a future white paper.

Neu1 Is Released From Activated Microglia, Stimulating Microglial Phagocytosis and Sensitizing Neurons to Glutamate

Neuraminidase 1 (Neu1) hydrolyses terminal sialic acid residues from glycoproteins and glycolipids, and is normally located in lysosomes, but can be released onto the surface of activated myeloid cells and microglia. We report that endotoxin/lipopolysaccharide-activated microglia released Neu1 into culture medium, and knockdown of Neu1 in microglia reduced both Neu1 protein and neuraminidase activity in the culture medium. Release of Neu1 was reduced by inhibitors of lysosomal exocytosis, and accompanied by other lysosomal proteins, including protective protein/cathepsin A, known to keep Neu1 active. Extracellular neuraminidase or over-expression of Neu1 increased microglial phagocytosis, while knockdown of Neu1 decreased phagocytosis. Microglial activation caused desialylation of microglial phagocytic receptors Trem2 and MerTK, and increased binding to Trem2 ligand galectin-3. Culture media from activated microglia contained Neu1, and when incubated with neurons induced their desialylation, and increased the neuronal death induced by low levels of glutamate. Direct desialylation of neurons by adding sialidase or inhibiting sialyltransferases also increased glutamate-induced neuronal death. We conclude that activated microglia can release active Neu1, possibly by lysosomal exocytosis, and this can both increase microglial phagocytosis and sensitize neurons to glutamate, thus potentiating neuronal death.

Brain Cells Release Calreticulin That Attracts and Activates Microglia, and Inhibits Amyloid Beta Aggregation and Neurotoxicity

Calreticulin is a chaperone, normally found in the endoplasmic reticulum, but can be released by macrophages into the extracellular medium. It is also found in cerebrospinal fluid bound to amyloid beta (Aβ). We investigated whether brain cells release calreticulin, and whether extracellular calreticulin had any effects on microglia and neurons relevant to neuroinflammation and neurodegeneration. We found that microglia release nanomolar levels of calreticulin when inflammatory-activated with lipopolysaccharide, when endoplasmic reticulum stress was induced by tunicamycin, or when cell death was induced by staurosporine, and that neurons release calreticulin when crushed. Addition of nanomolar levels of extracellular calreticulin was found to chemoattract microglia, and activate microglia to release cytokines TNF-α, IL-6 and IL-1β, as well as chemokine (C-C motif) ligand 2. Calreticulin blocked Aβ fibrillization and modified Aβ oligomerization, as measured by thioflavin T fluorescence and transmission electron microscopy. Extracellular calreticulin also altered microglial morphology and proliferation, and prevented Aβ-induced neuronal loss in primary neuron-glial cultures. Thus, calreticulin is released by microglia and neurons, and acts: as an alarmin to recruit and activate microglia, as an extracellular chaperone to prevent Aβ aggregation, and as a neuroprotectant against Aβ neurotoxicity.

Knockout of the P2Y6 Receptor Prevents Peri-Infarct Neuronal Loss after Transient, Focal Ischemia in Mouse Brain

After stroke, there is a delayed neuronal loss in brain areas surrounding the infarct, which may in part be mediated by microglial phagocytosis of stressed neurons. Microglial phagocytosis of stressed or damaged neurons can be mediated by UDP released from stressed neurons activating the P2Y₆ receptor on microglia, inducing microglial phagocytosis of such neurons. We show evidence here from a small trial that the knockout of the P2Y₆ receptor, required for microglial phagocytosis of neurons, prevents the delayed neuronal loss after transient, focal brain ischemia induced by endothelin-1 injection in mice. Wild-type mice had neuronal loss and neuronal nuclear material within microglia in peri-infarct areas. P2Y₆ receptor knockout mice had no significant neuronal loss in peri-infarct brain areas seven days after brain ischemia. Thus, delayed neuronal loss after stroke may in part be mediated by microglial phagocytosis of stressed neurons, and the P2Y₆ receptor is a potential treatment target to prevent peri-infarct neuronal loss.

Does Soluble TREM2 Protect Against Alzheimer's Disease?

Triggering Receptor Expressed in Myeloid Cells 2 (TREM2) is a pattern recognition receptor on myeloid cells, and is upregulated on microglia surrounding amyloid plaques in Alzheimer's disease (AD). Rare, heterozygous mutations in TREM2 (e.g., R47H) increase AD risk several fold. TREM2 can be cleaved at the plasma membrane by metalloproteases to release the ectodomain as soluble TREM2 (sTREM2). Wild-type sTREM2 binds oligomeric amyloid beta (Aβ) and acts as an extracellular chaperone, blocking and reversing Aβ oligomerization and fibrillization, and preventing Aβ-induced neuronal loss in vitro. Whereas, R47H sTREM2 increases Aβ fibrillization and neurotoxicity. AD brains expressing R47H TREM2 have more fibrous plaques with more neuritic pathology around these plaques, consistent with R47H sTREM2 promoting Aβ fibrillization relative to WT sTREM2. Brain expression or injection of wild-type sTREM2 reduces pathology in amyloid models of AD in mice, indicating that wild-type sTREM2 is protective against amyloid pathology. Levels of sTREM2 in cerebrospinal fluid (CSF) fall prior to AD, rise in early AD, and fall again in late AD. People with higher sTREM2 levels in CSF progress more slowly into and through AD than do people with lower sTREM2 levels, suggesting that sTREM2 protects against AD. However, some of these experiments can be interpreted as full-length TREM2 protecting rather than sTREM2, and to distinguish between these two possibilities, we need more experiments testing whether sTREM2 itself protects in AD and AD models, and at what stage of disease. If sTREM2 is protective, then treatments could be designed to elevate sTREM2 in AD.

The microglial P2Y6 receptor mediates neuronal loss and memory deficits in neurodegeneration

Microglia are implicated in neurodegeneration, potentially by phagocytosing neurons, but it is unclear how to block the detrimental effects of microglia while preserving their beneficial roles. The microglial P2Y₆ receptor (P2Y₆R) - activated by extracellular UDP released by stressed neurons - is required for microglial phagocytosis of neurons. We show here that injection of amyloid beta (Aβ) into mouse brain induces microglial phagocytosis of neurons, followed by neuronal and memory loss, and this is all prevented by knockout of P2Y₆R. In a chronic tau model of neurodegeneration (P301S TAU mice), P2Y₆R knockout prevented TAU-induced neuronal and memory loss. In vitro, P2Y₆R knockout blocked microglial phagocytosis of live but not dead targets and reduced tau-, Aβ-, and UDP-induced neuronal loss in glial-neuronal cultures. Thus, the P2Y₆ receptor appears to mediate Aβ- and tau-induced neuronal and memory loss via microglial phagocytosis of neurons, suggesting that blocking this receptor may be beneficial in the treatment of neurodegenerative diseases.

Neuronal Loss after Stroke Due to Microglial Phagocytosis of Stressed Neurons

After stroke, there is a rapid necrosis of all cells in the infarct, followed by a delayed loss of neurons both in brain areas surrounding the infarct, known as 'selective neuronal loss', and in brain areas remote from, but connected to, the infarct, known as 'secondary neurodegeneration'. Here we review evidence indicating that this delayed loss of neurons after stroke is mediated by the microglial phagocytosis of stressed neurons. After a stroke, neurons are stressed by ongoing ischemia, excitotoxicity and/or inflammation and are known to: (i) release "find-me" signals such as ATP, (ii) expose "eat-me" signals such as phosphatidylserine, and (iii) bind to opsonins, such as complement components C1q and C3b, inducing microglia to phagocytose such neurons. Blocking these factors on neurons, or their phagocytic receptors on microglia, can prevent delayed neuronal loss and behavioral deficits in rodent models of ischemic stroke. Phagocytic receptors on microglia may be attractive treatment targets to prevent delayed neuronal loss after stroke due to the microglial phagocytosis of stressed neurons.

Inflammatory neuronal loss in the substantia nigra induced by systemic lipopolysaccharide is prevented by knockout of the P2Y6 receptor in mice

Inflammation may contribute to multiple brain pathologies. One cause of inflammation is lipopolysaccharide/endotoxin (LPS), the levels of which are elevated in blood and/or brain during bacterial infections, gut dysfunction and neurodegenerative diseases, such as Parkinson’s disease. How inflammation causes neuronal loss is unclear, but one potential mechanism is microglial phagocytosis of neurons, which is dependent on the microglial P2Y₆ receptor. We investigated here whether the P2Y₆ receptor was required for inflammatory neuronal loss. Intraperitoneal injection of LPS on 4 successive days resulted in specific loss of dopaminergic neurons (measured as cells staining with tyrosine hydroxylase or NeuN) in the substantia nigra of wild-type mice, but no neuronal loss in cortex or hippocampus. This supports the hypothesis that neuronal loss in Parkinson’s disease may be driven by peripheral LPS. By contrast, there was no LPS-induced neuronal loss in P2Y₆ receptor knockout mice. In vitro, LPS-induced microglial phagocytosis of cells was prevented by inhibition of the P2Y₆ receptor, and LPS-induced neuronal loss was reduced in mixed glial–neuronal cultures from P2Y₆ receptor knockout mice. This supports the hypothesis that microglial phagocytosis contributes to inflammatory neuronal loss, and can be prevented by blocking the P2Y₆ receptor, suggesting that P2Y₆ receptor antagonists might be used to prevent inflammatory neuronal loss in Parkinson’s disease and other brain pathologies involving inflammatory neuronal loss.

The Phagocytic Code Regulating Phagocytosis of Mammalian Cells

Mammalian phagocytes can phagocytose (i.e. eat) other mammalian cells in the body if they display certain signals, and this phagocytosis plays fundamental roles in development, cell turnover, tissue homeostasis and disease prevention. To phagocytose the correct cells, phagocytes must discriminate which cells to eat using a 'phagocytic code' - a set of over 50 known phagocytic signals determining whether a cell is eaten or not - comprising find-me signals, eat-me signals, don't-eat-me signals and opsonins. Most opsonins require binding to eat-me signals - for example, the opsonins galectin-3, calreticulin and C1q bind asialoglycan eat-me signals on target cells - to induce phagocytosis. Some proteins act as 'self-opsonins', while others are 'negative opsonins' or 'phagocyte suppressants', inhibiting phagocytosis. We review known phagocytic signals here, both established and novel, and how they integrate to regulate phagocytosis of several mammalian targets - including excess cells in development, senescent and aged cells, infected cells, cancer cells, dead or dying cells, cell debris and neuronal synapses. Understanding the phagocytic code, and how it goes wrong, may enable novel therapies for multiple pathologies with too much or too little phagocytosis, such as: infectious disease, cancer, neurodegeneration, psychiatric disease, cardiovascular disease, ageing and auto-immune disease.

Microglial phagocytosis of neurons in neurodegeneration, and its regulation

There is growing evidence that excessive microglial phagocytosis of neurons and synapses contributes to multiple brain pathologies. RNA-seq and genome-wide association (GWAS) studies have linked multiple phagocytic genes to neurodegenerative diseases, and knock-out of phagocytic genes has been found to protect against neurodegeneration in animal models, suggesting that excessive microglial phagocytosis contributes to neurodegeneration. Here, we review recent evidence that microglial phagocytosis of live neurons and synapses causes neurodegeneration in animal models of Alzheimer's disease and other tauopathies, Parkinson's disease, frontotemporal dementias, multiple sclerosis, retinal degeneration and neurodegeneration induced by ischaemia, infection or ageing. We also review factors regulating microglial phagocytosis of neurons, including: nucleotides, frackalkine, phosphatidylserine, calreticulin, UDP, CD47, sialylation, complement, galectin-3, Apolipoprotein E, phagocytic receptors, Siglec receptors, cytokines, microglial epigenetics and expression profile. Some of these factors may be potential treatment targets to prevent neurodegeneration mediated by excessive microglial phagocytosis of live neurons and synapses.

Wild-type sTREM2 blocks Aβ aggregation and neurotoxicity, but the Alzheimer's R47H mutant increases Aβ aggregation

TREM2 is a pattern recognition receptor, expressed on microglia and myeloid cells, detecting lipids and Aβ and inducing an innate immune response. Missense mutations (e.g., R47H) of TREM2 increase risk of Alzheimer's disease (AD). The soluble ectodomain of wild-type TREM2 (sTREM2) has been shown to protect against AD in vivo, but the underlying mechanisms are unclear. We show that Aβ oligomers bind to cellular TREM2, inducing shedding of the sTREM2 domain. Wild-type sTREM2 bound to Aβ oligomers (measured by single-molecule imaging, dot blots, and Bio-Layer Interferometry) inhibited Aβ oligomerization and disaggregated preformed Aβ oligomers and protofibrils (measured by transmission electron microscopy, dot blots, and size-exclusion chromatography). Wild-type sTREM2 also inhibited Aβ fibrillization (measured by imaging and thioflavin T fluorescence) and blocked Aβ-induced neurotoxicity (measured by permeabilization of artificial membranes and by loss of neurons in primary neuronal-glial cocultures). In contrast, the R47H AD-risk variant of sTREM2 is less able to bind and disaggregate oligomeric Aβ but rather promotes Aβ protofibril formation and neurotoxicity. Thus, in addition to inducing an immune response, wild-type TREM2 may protect against amyloid pathology by the Aβ-induced release of sTREM2, which blocks Aβ aggregation and neurotoxicity. In contrast, R47H sTREM2 promotes Aβ aggregation into protofibril that may be toxic to neurons. These findings may explain how wild-type sTREM2 apparently protects against AD in vivo and why a single copy of the R47H variant gene is associated with increased AD risk.

Sialylation acts as a checkpoint for innate immune responses in the central nervous system

Sialic acids are monosaccharides that normally terminate the glycan chains of cell surface glyco-proteins and -lipids in mammals, and are highly enriched in the central nervous tissue. Sialic acids are conjugated to proteins and lipids (termed "sialylation") by specific sialyltransferases, and are removed ("desialylation") by neuraminidases. Cell surface sialic acids are sensed by complement factor H (FH) to inhibit complement activation or by sialic acid-binding immunoglobulin-like lectin (SIGLEC) receptors to inhibit microglial activation, phagocytosis, and oxidative burst. In contrast, desialylation of cells enables binding of the opsonins C1, calreticulin, galectin-3, and collectins, stimulating phagocytosis of such cells. Hypersialylation is used by bacteria and cancers as camouflage to escape immune recognition, while polysialylation of neurons protects synapses and neurogenesis. Insufficient lysosomal cleavage of sialylated molecules can lead to lysosomal accumulation of lipids and aggregated proteins, which if excessive may be expelled into the extracellular space. On the other hand, desialylation of immune receptors can activate them or trigger removal of proteins. Loss of inhibitory SIGLECs or FH triggers reduced clearance of aggregates, oxidative brain damage and complement-mediated retinal damage. Thus, cell surface sialylation recognized by FH, SIGLEC, and other immune-related receptors acts as a major checkpoint inhibitor of innate immune responses in the central nervous system, while excessive cleavage of sialic acid residues and consequently removing this checkpoint inhibitor may trigger lipid accumulation, protein aggregation, inflammation, and neurodegeneration.

Sialylation and Galectin-3 in Microglia-Mediated Neuroinflammation and Neurodegeneration

Microglia are brain macrophages that mediate neuroinflammation and contribute to and protect against neurodegeneration. The terminal sugar residue of all glycoproteins and glycolipids on the surface of mammalian cells is normally sialic acid, and addition of this negatively charged residue is known as “sialylation,” whereas removal by sialidases is known as “desialylation.” High sialylation of the neuronal cell surface inhibits microglial phagocytosis of such neurons, via: (i) activating sialic acid receptors (Siglecs) on microglia that inhibit phagocytosis and (ii) inhibiting binding of opsonins C1q, C3, and galectin-3. Microglial sialylation inhibits inflammatory activation of microglia via: (i) activating Siglec receptors CD22 and CD33 on microglia that inhibit phagocytosis and (ii) inhibiting Toll-like receptor 4 (TLR4), complement receptor 3 (CR3), and other microglial receptors. When activated, microglia release a sialidase activity that desialylates both microglia and neurons, activating the microglia and rendering the neurons susceptible to phagocytosis. Activated microglia also release galectin-3 (Gal-3), which: (i) further activates microglia via binding to TLR4 and TREM2, (ii) binds to desialylated neurons opsonizing them for phagocytosis via Mer tyrosine kinase, and (iii) promotes Aβ aggregation and toxicity in vivo. Gal-3 and desialylation may increase in a variety of brain pathologies. Thus, Gal-3 and sialidases are potential treatment targets to prevent neuroinflammation and neurodegeneration.

Mechanisms of cell death induced by arginase and asparaginase in precursor B-cell lymphoblasts

Arginase has therapeutic potential as a cytotoxic agent in some cancers, but this is unclear for precursor B acute lymphoblastic leukaemia (pre-B ALL), the commonest form of childhood leukaemia. We compared arginase cytotoxicity with asparaginase, currently used in pre-B ALL treatment, and characterised the forms of cell death induced in a pre-B ALL cell line 697. Arginase and asparaginase both efficiently killed 697 cells and mature B lymphoma cell line Ramos, but neither enzyme killed normal lymphocytes. Arginase depleted cellular arginine, and arginase-treated media induced cell death, blocked by addition of arginine or arginine-precursor citrulline. Asparaginase depleted both asparagine and glutamine, and asparaginase-treated media induced cell death, blocked by asparagine, but not glutamine. Both enzymes induced caspase cleavage and activation, chromatin condensation and phosphatidylserine exposure, indicating apoptosis. Both arginase- and asparaginase-induced death were blocked by caspase inhibitors, but with different sensitivities. BCL-2 overexpression inhibited arginase- and asparaginase-induced cell death, but did not prevent arginase-induced cytostasis, indicating a different mechanism of growth arrest. An autophagy inhibitor, chloroquine, had no effect on the cell death induced by arginase, but doubled the cell death induced by asparaginase. In conclusion, arginase causes death of lymphoblasts by arginine-depletion induced apoptosis, via mechanism distinct from asparaginase. Therapeutic implications for childhood ALL include: arginase might be used as treatment (but antagonised by dietary arginine and citrulline), chloroquine may enhance efficacy of asparaginase treatment, and partial resistance to arginase and asparaginase may develop by BCL-2 expression. Arginase or asparaginase might potentially be used to treat Burkitt lymphoma.

Extracellular tau induces microglial phagocytosis of living neurons in cell cultures

Tau is a microtubule-associated protein, found at high levels in neurons, and its aggregation is associated with neurodegeneration. Recently, it was found that tau can be actively secreted from neurons, but the effects of extracellular tau on neuronal viability are unclear. In this study, we investigated whether extracellular tau^2N4R can cause neurotoxicity in primary cultures of rat brain neurons and glial cells. Cell cultures were examined for neuronal loss, death, and phosphatidylserine exposure, as well as for microglial phagocytosis by fluorescence microscopy. Aggregation of tau^2N4R was assessed by atomic force microscopy. We found that extracellular addition of tau induced a gradual loss of neurons over 1-2 days, without neuronal necrosis or apoptosis, but accompanied by proliferation of microglia in the neuronal-glial co-cultures. Tau addition caused exposure of the 'eat-me' signal phosphatidylserine on the surface of living neurons, and this was prevented by elimination of the microglia or by inhibition of neutral sphingomyelinase. Tau also increased the phagocytic activity of pure microglia, and this was blocked by inhibitors of neutral sphingomyelinase or protein kinase C. The neuronal loss induced by tau was prevented by inhibitors of neutral sphingomyelinase, protein kinase C or the phagocytic receptor MerTK, or by eliminating microglia from the cultures. The data suggest that extracellular tau induces primary phagocytosis of stressed neurons by activated microglia, and identifies multiple ways in which the neuronal loss induced by tau can be prevented.

Activated microglia desialylate their surface, stimulating complement receptor 3-mediated phagocytosis of neurons

The glycoproteins and glycolipids of the cell surface have sugar chains that normally terminate in a sialic acid residue, but inflammatory activation of myeloid cells can cause sialidase enzymes to remove these residues, resulting in desialylation and altered activity of surface receptors, such as the phagocytic complement receptor 3 (CR3). We found that activation of microglia with lipopolysaccharide (LPS), fibrillar amyloid beta (Aβ), Tau or phorbol myristate acetate resulted in increased surface sialidase activity and desialylation of the microglial surface. Desialylation of microglia by adding sialidase, stimulated microglial phagocytosis of beads, but this was prevented by siRNA knockdown of CD11b or a blocking antibody to CD11b (a component of CR3). Desialylation of microglia by a sialyl-transferase inhibitor (3FAx-peracetyl-Neu5Ac) also stimulated microglial phagocytosis of beads. Desialylation of primary glial-neuronal co-cultures by adding sialidase or the sialyl-transferase inhibitor resulted in neuronal loss that was prevented by inhibiting phagocytosis with cytochalasin D or the blocking antibody to CD11b. Adding desialylated microglia to glial-neuronal cultures, in the absence of neuronal desialylation, also caused neuronal loss prevented by CD11b blocking antibody. Adding LPS or Aβ to primary glial-neuronal co-cultures caused neuronal loss, and this was prevented by inhibiting endogenous sialidase activity with N-acetyl-2,3-dehydro-2-deoxyneuraminic acid or blockage of CD11b. Thus, activated microglia release a sialidase activity that desialylates the cell surface, stimulating CR3-mediated phagocytosis of neurons, making extracellular sialidase and CR3 potential treatment targets to prevent inflammatory loss of neurons.

Calreticulin and Galectin-3 Opsonise Bacteria for Phagocytosis by Microglia

Opsonins are soluble, extracellular proteins, released by activated immune cells, and when bound to a target cell, can induce phagocytes to phagocytose the target cell. There are three known classes of opsonin: antibodies, complement factors and secreted pattern recognition receptors, but these have limited access to the brain. We identify here two novel opsonins of bacteria, calreticulin, and galectin-3 (both lectins that can bind lipopolysaccharide), which were released by microglia (brain-resident macrophages) when activated by bacterial lipopolysaccharide. Calreticulin and galectin-3 both bound to Escherichia coli, and when bound increased phagocytosis of these bacteria by microglia. Furthermore, lipopolysaccharide-induced microglial phagocytosis of E. coli bacteria was partially inhibited by: sugars, an anti-calreticulin antibody, a blocker of the calreticulin phagocytic receptor LRP1, a blocker of the galectin-3 phagocytic receptor MerTK, or simply removing factors released from the microglia, indicating this phagocytosis is dependent on extracellular calreticulin and galectin-3. Thus, calreticulin and galectin-3 are opsonins, released by activated microglia to promote clearance of bacteria. This innate immune response of microglia may help clear bacterial infections of the brain.

The endotoxin hypothesis of neurodegeneration

The endotoxin hypothesis of neurodegeneration is the hypothesis that endotoxin causes or contributes to neurodegeneration. Endotoxin is a lipopolysaccharide (LPS), constituting much of the outer membrane of gram-negative bacteria, present at high concentrations in gut, gums and skin and in other tissue during bacterial infection. Blood plasma levels of endotoxin are normally low, but are elevated during infections, gut inflammation, gum disease and neurodegenerative disease. Adding endotoxin at such levels to blood of healthy humans induces systemic inflammation and brain microglial activation. Adding high levels of endotoxin to the blood or body of rodents induces microglial activation, priming and/or tolerance, memory deficits and loss of brain synapses and neurons. Endotoxin promotes amyloid β and tau aggregation and neuropathology, suggesting the possibility that endotoxin synergises with different aggregable proteins to give different neurodegenerative diseases. Blood and brain endotoxin levels are elevated in Alzheimer's disease, which is accelerated by systemic infections, including gum disease. Endotoxin binds directly to APOE, and the APOE4 variant both sensitises to endotoxin and predisposes to Alzheimer's disease. Intestinal permeability increases early in Parkinson's disease, and injection of endotoxin into mice induces α-synuclein production and aggregation, as well as loss of dopaminergic neurons in the substantia nigra. The gut microbiome changes in Parkinson's disease, and changing the endotoxin-producing bacterial species can affect the disease in patients and mouse models. Blood endotoxin is elevated in amyotrophic lateral sclerosis, and endotoxin promotes TDP-43 aggregation and neuropathology. Peripheral diseases that elevate blood endotoxin, such as sepsis, AIDS and liver failure, also result in neurodegeneration. Endotoxin directly and indirectly activates microglia that damage neurons via nitric oxide, oxidants and cytokines, and by phagocytosis of synapses and neurons. The endotoxin hypothesis is unproven, but if correct, then neurodegeneration may be reduced by decreasing endotoxin levels or endotoxin-induced neuroinflammation.

Neuronal Cell Death

Neuronal cell death occurs extensively during development and pathology, where it is especially important because of the limited capacity of adult neurons to proliferate or be replaced. The concept of cell death used to be simple as there were just two or three types, so we just had to work out which type was involved in our particular pathology and then block it. However, we now know that there are at least a dozen ways for neurons to die, that blocking a particular mechanism of cell death may not prevent the cell from dying, and that non-neuronal cells also contribute to neuronal death. We review here the mechanisms of neuronal death by intrinsic and extrinsic apoptosis, oncosis, necroptosis, parthanatos, ferroptosis, sarmoptosis, autophagic cell death, autosis, autolysis, paraptosis, pyroptosis, phagoptosis, and mitochondrial permeability transition. We next explore the mechanisms of neuronal death during development, and those induced by axotomy, aberrant cell-cycle reentry, glutamate (excitoxicity and oxytosis), loss of connected neurons, aggregated proteins and the unfolded protein response, oxidants, inflammation, and microglia. We then reassess which forms of cell death occur in stroke and Alzheimer's disease, two of the most important pathologies involving neuronal cell death. We also discuss why it has been so difficult to pinpoint the type of neuronal death involved, if and why the mechanism of neuronal death matters, the molecular overlap and interplay between death subroutines, and the therapeutic implications of these multiple overlapping forms of neuronal death.

Neurophagy, the phagocytosis of live neurons and synapses by glia, contributes to brain development and disease

It was previously thought that neurons were phagocytosed only when dead or dying. However, it is increasingly clear that viable synapses, dendrites, axons and whole neurons can be phagocytosed alive (defined here as neurophagy), and this may contribute to a wide range of developmental, physiological and pathological processes. Phagocytosis of live synapses, dendrites and axons by glia contributes to experience-dependent sculpting of neuronal networks during development, but excessive phagocytosis of synapses may contribute to pathology in Alzheimer's disease, schizophrenia and ageing. Neurons can expose phosphatidylserine or calreticulin, which act as 'eat me' signals provoking phagocytosis via microglial receptors, whereas sialylation of neuronal surfaces acts as a 'don't eat me' signal that inhibits phagocytosis and desialylation can provoke phagocytosis. Opsonins, such as complement components and apolipoproteins, are released during inflammation and enhance engulfment. Phagocytosis of neurons is seen in multiple human diseases, but it is as yet unclear whether inhibition of phagocytosis will be beneficial in treating neurological diseases. Here we review the signals regulating glial phagocytosis of live neurons and synapses, and the involvement of this phagocytosis in development and disease.

Deciphering microglial diversity in Alzheimer's disease

A subset of microglia appear primed to protect against AD neurodegeneration

Anti-CD47 antibodies induce phagocytosis of live, malignant B cells by macrophages via the Fc domain, resulting in cell death by phagoptosis

When expressed on the surface of cells, CD47 inhibits phagocytosis of these cells by phagocytes. Most human cancers overexpress CD47, and antibodies to CD47 have shown a remarkable ability to clear a range of cancers in animal models. However, the mechanism by which these antibodies cause cancer cell death is unclear. We find that CD47 is expressed on the surface of three B-cell lines from human malignancies: 697 (pre-B-ALL lymphoblasts), Ramos and DG-75 (both mature B-cells, Burkitt’s lymphoma), and anti-CD47 antibodies greatly increase the phagocytosis of all three cell line by macrophages. In the presence of macrophages, the antibodies cause clearance of the lymphoblasts within hours, but in the absence of macrophages, the antibodies have no effect on lymphoblast viability. Macrophages engulf viable lymphoblasts containing mitochondria with a normal membrane potential, but following engulfment the mitochondrial membrane potential is lost indicating a loss of viability. Inhibition of phagocytosis protects lymphoblasts from death indicating that phagocytosis is required for anti-CD47 mediated cell death. Blocking either the antibody Fc domain or Fc receptors inhibits antibody-induced phagocytosis. Antibodies against cell surface markers CD10 or CD19 also induced Fc-domain-dependent phagocytosis, but at a lower level commensurate with expression. Thus, phagoptosis may contribute to the efficacy of a number of therapeutic antibodies used in cancer therapy, as well as potentially endogenous antibodies. We conclude that anti-CD47 antibodies induce phagocytosis by binding CD47 on lymphoblast and Fc receptors on macrophages, resulting in cell death by phagocytosis, i.e. phagoptosis.

Activated Microglia Desialylate and Phagocytose Cells via Neuraminidase, Galectin-3, and Mer Tyrosine Kinase

Activated microglia can phagocytose dying, stressed, or excess neurons and synapses via the phagocytic receptor Mer tyrosine kinase (MerTK). Galectin-3 (Gal-3) can cross-link surface glycoproteins by binding galactose residues that are normally hidden below terminal sialic acid residues. Gal-3 was recently reported to opsonize cells via activating MerTK. We found that LPS-activated BV-2 microglia rapidly released Gal-3, which was blocked by calcineurin inhibitors. Gal-3 bound to MerTK on microglia and to stressed PC12 (neuron-like) cells, and it increased microglial phagocytosis of PC12 cells or primary neurons, which was blocked by inhibition of MerTK. LPS-activated microglia exhibited a sialidase activity that desialylated PC12 cells and could be inhibited by Tamiflu, a neuraminidase (sialidase) inhibitor. Sialidase treatment of PC12 cells enabled Gal-3 to bind and opsonize the live cells for phagocytosis by microglia. LPS-induced microglial phagocytosis of PC12 was prevented by small interfering RNA knockdown of Gal-3 in microglia, lactose inhibition of Gal-3 binding, inhibition of neuraminidase with Tamiflu, or inhibition of MerTK by UNC569. LPS-induced phagocytosis of primary neurons by primary microglia was also blocked by inhibition of MerTK. We conclude that activated microglia release Gal-3 and a neuraminidase that desialylates microglial and PC12 surfaces, enabling Gal-3 binding to PC12 cells and their phagocytosis via MerTK. Thus, Gal-3 acts as an opsonin of desialylated surfaces, and inflammatory loss of neurons or synapses may potentially be blocked by inhibiting neuraminidases, Gal-3, or MerTK.

Activated microglia cause reversible apoptosis of pheochromocytoma cells, inducing their cell death by phagocytosis

Some apoptotic processes, such as phosphatidylserine exposure, are potentially reversible and do not necessarily lead to cell death. However, phosphatidylserine exposure can induce phagocytosis of a cell, resulting in cell death by phagocytosis: phagoptosis. Phagoptosis of neurons by microglia might contribute to neuropathology, whereas phagoptosis of tumour cells by macrophages might limit cancer. Here, we examined the mechanisms by which BV-2 microglia killed co-cultured pheochromocytoma (PC12) cells that were either undifferentiated or differentiated into neuronal cells. We found that microglia activated by lipopolysaccharide rapidly phagocytosed PC12 cells. Activated microglia caused reversible phosphatidylserine exposure on and reversible caspase activation in PC12 cells, and caspase inhibition prevented phosphatidylserine exposur and decreased subsequent phagocytosis. Nitric oxide was necessary and sufficient to induce the reversible phosphatidylserine exposure and phagocytosis. The PC12 cells were not dead at the time they were phagocytised, and inhibition of their phagocytosis left viable cells. Cell loss was inhibited by blocking phagocytosis mediated by phosphatidylserine, MFG-E8, vitronectin receptors or P2Y6 receptors. Thus, activated microglia can induce reversible apoptosis of target cells, which is insufficient to cause apoptotic cell death, but sufficient to induce their phagocytosis and therefore cell death by phagoptosis.

How microglia kill neurons

Microglia are resident brain macrophages that become inflammatory activated in most brain pathologies. Microglia normally protect neurons, but may accidentally kill neurons when attempting to limit infections or damage, and this may be more common with degenerative disease as there was no significant selection pressure on the aged brain in the past. A number of mechanisms by which activated microglia kill neurons have been identified, including: (i) stimulation of the phagocyte NADPH oxidase (PHOX) to produce superoxide and derivative oxidants, (ii) expression of inducible nitric oxide synthase (iNOS) producing NO and derivative oxidants, (iii) release of glutamate and glutaminase, (iv) release of TNFα, (v) release of cathepsin B, (vi) phagocytosis of stressed neurons, and (vii) decreased release of nutritive BDNF and IGF-1. PHOX stimulation contributes to microglial activation, but is not directly neurotoxic unless NO is present. NO is normally neuroprotective, but can react with superoxide to produce neurotoxic peroxynitrite, or in the presence of hypoxia inhibit mitochondrial respiration. Glutamate can be released by glia or neurons, but is neurotoxic only if the neurons are depolarised, for example as a result of mitochondrial inhibition. TNFα is normally neuroprotective, but can become toxic if caspase-8 or NF-κB activation are inhibited. If the above mechanisms do not kill neurons, they may still stress the neurons sufficiently to make them susceptible to phagocytosis by activated microglia. We review here whether microglial killing of neurons is an artefact, makes evolutionary sense or contributes in common neuropathologies and by what mechanisms. This article is part of a Special Issue entitled SI: Neuroprotection.

Neuroinflammation in Alzheimer's disease

Increasing evidence suggests that Alzheimer's disease pathogenesis is not restricted to the neuronal compartment, but includes strong interactions with immunological mechanisms in the brain. Misfolded and aggregated proteins bind to pattern recognition receptors on microglia and astroglia, and trigger an innate immune response characterised by release of inflammatory mediators, which contribute to disease progression and severity. Genome-wide analysis suggests that several genes that increase the risk for sporadic Alzheimer's disease encode factors that regulate glial clearance of misfolded proteins and the inflammatory reaction. External factors, including systemic inflammation and obesity, are likely to interfere with immunological processes of the brain and further promote disease progression. Modulation of risk factors and targeting of these immune mechanisms could lead to future therapeutic or preventive strategies for Alzheimer's disease.

Tumour necrosis factor alpha-induced neuronal loss is mediated by microglial phagocytosis

•

TNF-α induces neuronal loss in culture without neuronal necrosis or apoptosis.

•

TNF-α induces microglial phagocytosis, and the neuronal loss requires microglia.

•

Neuronal loss requires the microglial vitronectin receptor, P2Y₆ receptor and MFG-E8.

•

Blocking these phagocytic receptors leaves viable neurons.

•

TNF-α induced phagoptosis might contribute to neuroinflammatory pathologies.

Tumour necrosis factor-α (TNF-α) is a pro-inflammatory cytokine, expressed in many brain pathologies and associated with neuronal loss. We show here that addition of TNF-α to neuronal–glial co-cultures increases microglial proliferation and phagocytosis, and results in neuronal loss that is prevented by eliminating microglia. Blocking microglial phagocytosis by inhibiting phagocytic vitronectin and P2Y₆ receptors, or genetically removing opsonin MFG-E8, prevented TNF-α induced loss of live neurons. Thus TNF-α appears to induce neuronal loss via microglial activation and phagocytosis of neurons, causing neuronal death by phagoptosis.

Inhibition of UDP/P2Y6 purinergic signaling prevents phagocytosis of viable neurons by activated microglia in vitro and in vivo

Microglia activated through Toll-like receptor (TLR)-2 or -4 can cause neuronal death by phagocytosing otherwise-viable neurons—a form of cell death called “phagoptosis.” UDP release from neurons has been shown to provoke microglial phagocytosis of neurons via microglial P2Y₆ receptors, but whether inhibition of this process affects neuronal survival is unknown. We tested here whether inhibition of P2Y₆ signaling could prevent neuronal death in inflammatory conditions, and whether UDP signaling can induce phagoptosis of stressed but viable neurons. We find that delayed neuronal loss and death in mixed neuronal/glial cultures induced by the TLR ligands lipopolysaccharide (LPS) or lipoteichoic acid was prevented by: apyrase (to degrade nucleotides), Reactive Blue 2 (to inhibit purinergic signaling), or MRS2578 (to specifically block P2Y₆ receptors). In each case, inflammatory activation of microglia was not affected, and the rescued neurons remained viable for at least 7 days. Blocking P2Y₆ receptors with MRS2578 also prevented phagoptosis of neurons induced by 250 nM amyloid beta 1–42, 5 μM peroxynitrite, or 50 μM 3-morpholinosydnonimine (which releases reactive oxygen and nitrogen species). Furthermore, the P2Y₆ receptor agonist UDP by itself was sufficient to stimulate microglial phagocytosis and to induce rapid neuronal loss that was prevented by eliminating microglia or inhibiting phagocytosis. In vivo, injection of LPS into rat striatum induced microglial activation and delayed neuronal loss and blocking P2Y₆ receptors with MRS2578 prevented this neuronal loss. Thus, blocking UDP/P2Y₆ signaling is sufficient to prevent neuronal loss and death induced by a wide range of stimuli that activate microglial phagocytosis of neurons.

Spin density projection-assisted R2 magnetic resonance imaging of the liver in the management of body iron stores in patients receiving multiple red blood cell transfusions: an audit and retrospective study in South Australia

AIM

To assess the impact of non-invasive monitoring of liver iron concentration (LIC) on management of body iron stores in patients receiving multiple blood transfusions.

METHOD

A retrospective audit was conducted on clinical data from 40 consecutive subjects with haemolytic anaemias or ineffective haematopoiesis who had been monitored non-invasively for LIC over a period of at least 1 year. LIC was measured with spin density projection-assisted proton transverse relaxation rate-magnetic resonance imaging.

RESULTS

Nineteen clinical decisions were explicitly documented in the case notes as being based on LIC results. Decisions comprised initiation of chelation therapy, increasing chelator dose, decreasing chelator dose and change of mode of delivery of deferioxamine from subcutaneous to intravenous. The geometrical mean LIC for the cohort dropped significantly (P= 0.008) from 6.8 mg Fe/g dry tissue at initial measurement to 4.8 mg Fe/g dry tissue at final measurement. The proportion of subjects with LIC in the range associated with greatly increased risk of cardiac disease and death (>15 mg Fe/g dry tissue) dropped significantly (P= 0.01) from 14 of 40 subjects at initial measurement to 5 of 40 subjects at final measurement. No significant changes in the geometrical mean of serum ferritin or the proportion of subjects with serum ferritin above 2500 or 1500 µg/L were observed.

CONCLUSIONS

The data are consistent with previous observations that introduction of non-invasive monitoring of LIC can contribute to a decreased body iron burden through improved clinical decision making and improved feedback to patients and hence improved adherence to chelation therapy.

Deoxyglucose prevents neurodegeneration in culture by eliminating microglia

Background

2-Deoxy-d-glucose is an inhibitor of glycolysis, which is protective in animal models of brain pathology, but the mechanisms of this protection are unclear. We examined whether, when and how deoxyglucose protects neurons in co-culture with astrocytes and microglia. Microglia are brain macrophages, which can damage neurons in inflammatory conditions.

Methods

Deoxyglucose was added to primary cultures of microglia and astrocytes from rat cortex, or neurons and glia from rat cerebellum, or the BV-2 microglial cell line, and cell death and cell functions were evaluated.

Results

Surprisingly, addition of deoxyglucose induced microglial loss and prevented spontaneous neuronal loss in long-term cultures of neurons and glia, while elimination of microglia by l-leucine-methyl ester prevented the deoxyglucose-induced neuroprotection. Deoxyglucose also prevented neuronal loss induced by addition of amyloid beta or disrupted neurons (culture models of Alzheimer’s disease and brain trauma respectively). However, deoxyglucose greatly increased the neuronal death induced by hypoxia. Addition of deoxyglucose to pure microglia induced necrosis and loss, preceded by rapid ATP depletion and followed by phagocytosis of the microglia. Deoxyglucose did not kill astrocytes or neurons.

Conclusions

We conclude that deoxyglucose causes microglial loss by ATP depletion, and this can protect neurons from neurodegeneration, except in conditions of hypoxia. Deoxyglucose may thus be beneficial in brain pathologies mediated by microglia, including brain trauma, but not where hypoxia is involved.

Inflammation induces multinucleation of Microglia via PKC inhibition of cytokinesis, generating highly phagocytic multinucleated giant cells

Microglia are brain macrophages, which can undergo multinucleation to give rise to multinucleated giant cells that accumulate with ageing and some brain pathologies. However, the origin, regulation and function of multinucleate microglia remain unclear. We found that inflammatory stimuli, including lipopolysaccharide, amyloid β, α-synuclein, tumour necrosis factor-α and interferon γ, but not interleukin-4, induced multinucleation of cultured microglia: primary rat cortical microglia and the murine microglial cell line BV-2. Inflammation-induced multinucleation was prevented by a protein kinase C (PKC) inhibitor Gö6976 (100 nM) and replicated by a PKC activator phorbol myristate acetate (160 nM). Multinucleation was reversible and not because of cell fusion or phagocytosis, but rather failure of cytokinesis. Time-lapse imaging revealed that some dividing cells failed to abscise, even after formation of long cytoplasmic bridges, followed by retraction of bridge and reversal of cleavage furrow to form multinucleate cells. Multinucleate microglia were larger and 2-4 fold more likely to phagocytose large beads and both dead and live PC12 cells. We conclude that multinucleate microglia are reversibly generated by inflammation via PKC inhibition of cytokinesis, and may have specialized functions/dysfunctions including the phagocytosis of other cells. Inflammation resulted in the accumulation of multiple nuclei per cell in cultured microglia. This multinucleation was reversible and due to a PKC-dependent block of the last step of cell division. Multinucleate microglia were larger and had a greater capacity to phagocytose other cells, suggesting they might remove neurons in the brain.

Introduction: Focus on mitochondria

This is a series of papers (three reviews and five regular papers) focused on Mitochondria from authors contributing to two FEBS courses on Mitochondria held in 2012. These papers focus on: (i) mitochondrial biogenesis (regulation by phosphorylation, import of SOD1 and Mia40, mitophagy and tRNA variants) or (ii) mitochondrial pathology (ischaemia, obesity and Parkinson's disease).

Rotenone induces neuronal death by microglial phagocytosis of neurons

Rotenone, a common pesticide and inhibitor of mitochondrial complex I, induces microglial activation and loss of dopaminergic neurons in models of Parkinson's disease. However, the mechanisms of rotenone neurotoxicity are still poorly defined. Here, we used primary neuronal/glial cultures prepared from rat cerebella to investigate the contribution of microglia to neuronal cell death induced by low concentrations of rotenone. Rotenone at 2.5 nm induced neuronal loss over several days without increasing the numbers of necrotic or apoptotic neurons, and neuronal loss/death could be prevented by selective removal of microglia. Rotenone increased microglial proliferation and phagocytic activity, without increasing tumour necrosis factor-α release. Rotenone-induced neuronal loss/death could be prevented by inhibition of phagocytic signalling between neurons and microglia with: cyclo(Arg-Gly-Asp-d-Phe-Val) (to block the microglial vitronectin receptor); MRS2578 (to block the microglial P2Y6 receptor); or either annexin V or an antibody against phosphatidylserine (to block exposed phosphatidylserine, a well-characterized neuronal 'eat-me' signal). As inhibition of phagocytosis by five different means prevented neuronal loss without increasing neuronal death, these data indicate that rotenone neurotoxicity is at least partially mediated by microglial phagocytosis of otherwise viable neurons (phagoptosis). Thus, neuronal loss in Parkinson's disease and other neurological diseases might be prevented by blocking phagocytic signalling.

In the eye of the storm: mitochondrial damage during heart and brain ischaemia

We review research investigating mitochondrial damage during heart and brain ischaemia, focusing on the mechanisms and consequences of ischaemia-induced and/or reperfusion-induced: (a) inhibition of mitochondrial respiratory complex I; (b) release of cytochrome c from mitochondria; (c) changes to mitochondrial phospholipids; and (d) nitric oxide inhibition of mitochondria. Heart ischaemia causes inhibition of cytochrome oxidase and complex I, release of cytochrome c, and induction of permeability transition and hydrolysis and oxidation of mitochondrial phospholipids, but some of the mechanisms are unclear. Brain ischaemia causes inhibition of complexes I and IV, but other effects are less clear.

Lactadherin/MFG-E8 is essential for microglia-mediated neuronal loss and phagoptosis induced by amyloid β

Nanomolar β-amyloid peptide (Aβ) can induce neuronal loss in culture by activating microglia to phagocytose neurons. We report here that this neuronal loss is mediated by the bridging protein lactadherin/milk-fat globule epidermal growth factor-like factor 8 (MFG-E8), which is released by Aβ-activated microglia, binds to co-cultured neurons and opsonizes neurons for phagocytosis by microglia. Aβ stimulated microglial phagocytosis, but did not opsonize neurons for phagocytosis. Aβ (250 nM) induced delayed neuronal loss in mixed glial-neuronal mouse cultures that required microglia and occurred without increasing neuronal apoptosis or necrosis. This neuronal death/loss was prevented by antibodies to MFG-E8 and was absent in cultures from Mfge8 knockout mice (leaving viable neurons), but was reconstituted by addition of recombinant MFG-E8. Thus, nanomolar Aβ caused neuronal death by inducing microglia to phagocytose otherwise viable neurons via MFG-E8. The direct neurotoxicity of micromolar Aβ was not affected by MFG-E8. The essential role of MFG-E8 in Aβ-induced phagoptosis, suggests this bridging protein as a potential therapeutic target to prevent neuronal loss in Alzheimer's disease.