Introduction

Microglia are small glial cells found in the brain and spinal cord which unlike classical neuroglia (astroglia and ependymal cells, oligodendrocytes) are mesodermal in origin. It is now accepted that microglial cells can be acutely blood-derived in the adult under certain pathological conditions (“bone-marrow derived microglia”, e.g. [198]). Microglial cells have the potential to develop into full-blown macrophages but their morphological and functional spectrum is highly regulated in vivo resulting in various intermediate states of activation. Adapted to their CNS environment, the functions of microglia differ from those of macrophages in other organs.

Twenty years have passed since we proposed the concept of the microglial “immune network” in the CNS [65]. At the time, a leading textbook of neuropathology questioned the very existence of microglial cells and suggested to abandon the term [45]. Obviously, this did not happen and more than 10,000 articles on microglia have been published since then. Microglia are now firmly established as a key cellular element in the CNS and they are recognized to serve as the brain and spinal cord’s innate immune system [168]. This article aims to provide a critical review of the most recent developments in the microglia literature dating back no more than 5 years in most cases.

Interactions with other cells

The view that glia merely represent non-excitable support cells has changed radically in recent years. Astrocytes are now seen as local communication elements within the CNS that can generate various signals, e.g. through the regulated release of ‘gliotransmitters’ including glutamate [187]. “Microgliotransmitters” are not known at present but astrocytes certainly represent a highly communicative partner of microglial cells. Therefore, it is reasonable to assume that astrocytic information-processing [4] can be influenced by microglial activation. This novel concept gives an additional, CNS specific functional meaning to a number of the phenomena discussed below. The reverse is certainly also true: astrocytes have an influence upon microglial behavior. For instance, astrocytes play a critical role in the activation of microglia under infectious conditions [140], and chemokine expression by astrocytes appears to be involved in microglia/macrophage activation in multiple sclerosis with MCP-1/CCL2 and IP-10/CXCL10 directing reactive gliosis [178]. Together, astrocytes and microglia differentially regulate trafficking of lymphocyte subsets across brain endothelial cells [79], and IL-10 production by microglia and astrocytes can be induced through pro-inflammatory cytokines released from leptomeningeal cells [196].

Microglia subtypes and microglia cellular markers

Brain regional specialization of glial cells as determined by anatomical constraints is a well known fact. Microglial cells also vary in terms of numerical density throughout the CNS. However, whether there are true subsets of microglia is a different matter. This was originally suggested in 1998 [12] but no substantial follow-up research has been performed. In contrast, functional subsets of microglial cells do make their appearance under pathological conditions [33]. For instance, in Alzheimer’s disease (AD) two microglial subtypes have been suggested to play different neuroimmunomodulating roles [160].

Microglial markers are crucial for the identification of microglial cells in tissue sections and much of the historical controversy surrounding them can be explained by a decades-long lack thereof. Today, we have a number of established markers for microglial cells in different species but a single specific molecular marker that does not also label peripheral macrophages is so far unknown and perhaps does not even exist [158]. Known microglia markers are cell-type specific in the sense that they do not label other glia or neurons. The comprehensive molecular characterization of microglia by means of high-throughput technologies such as microarrays [2, 23, 47, 57, 60, 129, 130, 147, 189] is in its infancy but a cell-type specific systems biological definition of microglia is likely to emerge. However, simple expression signatures of microglia may vary between cells from different brain regions or between individual cells, as functional and phenotypic heterogeneity of microglial cells has long been known to exist.

Perhaps the most versatile immunocytochemical marker for microglial cells is Iba1, which is detectable by an antibody directed against the EF hand protein which was developed in Shinichi Kohsaka’s laboratory [82, 84]. Monoclonal and polyclonal versions of this antibody are available which enable most useful double-labeling procedures [69, 171, 194].

More recently identified markers of microglia are GLUT5 [157], CD163 [19, 149], and the C-type lectin CD209b [141]. CCR2 [198] and nestin [176] have also been described in microglia as well as CD34 [5, 98]. Toll-like receptor 2 (TLR2) and Toll-like receptor 4 (TLR4) are suggested to be involved in the determination of microglia phenotype and function [104]. Detection of the neurotransmitter gamma-aminobutyric acid (GABA) and of glutamic acid decarboxylase (GAD-67) in microglia following long exposure (more than 10 days) to low doses (10 ng/ml) of the ‘proinflammatory’ T-cell derived cytokine, IFN-gamma [22] attests to their great functional plasticity, which is especially pronounced in vitro.

Microglial cell lines

One microglial cell line, BV-2 [14] has become particularly popular with researchers but reservations exist that it does not model primary microglia accurately [42, 78]. For instance, BV-2 cells did not give reliable results in chemotaxis assays, not only with chemokines but also with other well-known attractors of microglial cells such as ATP [42]. Horvath et al. [78] recommended that the use of this microglial cell line should be carefully considered. While there is a reasonable motivation to reduce the number of animal experiments [75], we feel that BV-2 should only be used for screening purposes with validation performed in primary cultures or, preferably, an established experimental in vivo setting. The same applies to other microglial cell lines since immortalization significantly affects a cell’s biology.

Recent findings

Developmental aspects

The role of microglia during normal development of the nervous system is not fully understood [25]. Monier et al. [126] studied the topographical distribution of microglia during development of the human diencephalon and telencephalon. At 5.5 gestational weeks (gw), the first intracerebral microglial cells were seen close to the meninges and choroid plexus near the di-telencephalic fissure [126]. The authors also reported that in the developing diencephalon, microglial clusters were located in junctional regions of the white matter anlagen, most notably at the junctions of the internal capsule with the thalamic projections, the external capsule, and the cerebral peduncle. At 10–12 gw, Iba1+/RCA-1+/CD68+/CD45+ cells accumulated in the cortical anlagen, forming a tangential band at the junction between the cortical plate and the subplate [126]. Subsequently, microglial clusters increased markedly in size and cellular density (10–16 gw). Beginning in the middle of the second trimester, microglial cells colonized the entire cerebral parenchyma, they developed a ramified morphology, and down-regulated their surface antigens, but remained more numerous in the white matter [126]. In another study using human brain tissue, the finding of transient, developmental stage-dependent overabundance of CD68-activated microglia in the cerebral white matter of fetuses was interpreted as potential “priming” of this area for brain insults such as periventricular leukomalacia, where microglial activation occurs in immature cerebral white matter secondary to hypoxia–ischemia and/or infection [13]. Checchin and co-workers [28] investigated whether microglia are implicated in retinal blood vessel formation in humans and rodents. They concluded that the traditional definition of microglia as “merely immunocompetent cells” should be reconsidered to encompass a new function related to blood vessel formation.

In a study in cats, the phenotype of microglial cells was found to depend on the postnatal visual experience, suggesting that microglia may interact with axons of visual neurons [150]. Normal microglia are not believed to play an active role in triggering apoptosis of developing motor neurons. Instead, they are thought to act as phagocytes for the removal of dying cells during the process of programmed cell death [25]. This may be different in fish [142]. It is worth noting that subventricular zone microglia possess a special capacity for massive in vitro expansion [118]. Another recent observation is that MHC class II expressing microglial cells are found in focal cortical dysplasia where they have been suggested to contribute to the epileptogenicity of such foci [16].

Functions of microglia in normal brain

More than a decade ago, Georg Kreutzberg coined the term “microglial sensor of pathology” [97] which captures the essence of microglial cell function and, not surprisingly, has become a widely accepted notion today. Clearly, microglia are involved in most if not all known CNS pathologies. However, strong emphasis on the role of microglia in disease conditions may distract from the fact that “resting microglia” are resident but not functionally silent cells. It seems that they have important roles to play in normal brain. For instance, Wake et al. [188] have shown recently that part of the dynamic motility of resting microglial processes in vivo is directed toward synapses and the authors propose that microglia vigilantly monitor and respond to the functional status of synapses. Their work also shows that microglia sense defunct synapses and suggests that the integrity of the CNS circuitry may be monitored and diagnosed by the “microglial sensor”. This work indicates a possible novel function of microglia as the CNS ‘electrician’ which deserves specific study. Wake et al.’s work further underlines the validity of the hypothesis posited above that microglia can potentially influence information processing in the CNS, either indirectly via their interactions with astrocytes, or directly by interacting with synapses.

Microglial cells possess alpha-amino-3-hydroxy-5-methylisoxazole-4-propionic acid (AMPA) and metabotropic glutamate receptors which mediate their directed chemotaxis [108]. It seems possible that this mechanism plays a role in “synaptic stripping” [15, 66, 182] including in human cortex (cf. Fig. 1). Constant surveillance of CNS tissue by microglial cells is considered to be of special relevance for the defense against viruses [180]. Viral encephalitis is typically accompanied by the formation of microglial rod cells (Fig. 1) which appear to form in cerebral cortex as a result of microglial-dendritic interactions.

Fig. 1
figure 1

MHC class II labeled microglia in human cortex (serologically confirmed SSPE). a and b demonstrate how selective the microglia “attack” on cortical neurons is (arrows point to unaffected nerve cells) suggesting that the phenomenon is reactive rather than aggressive. c, e Some nerve cells become completely “wrapped” by microglia while microglial rod cells form along dendrites (d). Occasional neuronophagia (f) is also observed whereas white matter shows individual activated microglial cells with stout processes (g). Scale bar 20 μm in a, d, g, and 40 μm in b, c, e, f

Full size image

Mechanisms of microglial activation

Activated microglia in tissue sections are identified based on combined morphological and immunophenotypic changes which distinguish them from their resting phenotype. These changes in microglial appearance provided the foundation for introducing the concept of microglia functional plasticity in the 1980s [170] and they remain relevant today. As mentioned, even after 20 years of intense research, there is still no single molecular marker that allows the unequivocal distinction between resting and activated microglia, although some rat antibodies have been claimed to fulfill that purpose. ED1 and OX-6 antibodies are examples of this, but both have their unique shortcomings: ED1 is a marker for phagocytic activity but does not label activated, non-phagocytic cells, and OX-6, while often upregulated on some activated cells does not get expressed by all of them and is also found on some resting microglia in normal brain. Thus, the distinction between activated/non-phagocytic and activated/phagocytic, which is embedded in the functional plasticity concept, has been confirmed using multiple markers in combination with morphological assessments. Ideally, such differences are validated in a paired experimental model [127] but trained observers are quite capable of recognizing even subtle states of microglial activation on the basis of their increased expression of marker molecules because the latter is usually combined with morphological alterations making the cells stand out against resting phenotypes. In recent years, the concept of microglia functional plasticity has been expanded with the recognition of a fourth microglial phenotype, termed dystrophic [173] (Figs. 2, 3). Conceptually, the addition of microglial dystrophy has widened our biological perspective in that microglial degeneration, in addition to activation, now becomes a consideration. As discussed here and elsewhere [169], microglial dystrophy is likely to be relevant especially in the context of aging and aging-related neurodegenerative diseases while primary gliodegeneration also exists [35]. Much needs to be learned about the processes that lead to microglial degeneration, as well as the consequences of microglial cell death for brain function. However, one interesting possibility is that chronic microglial activation if sustained over long periods could lead to microglial over-activation followed by microglial degeneration. Currently, there is in vitro evidence to support the idea that overactivation leads to microglial cell death [106, 143], and degenerating microglia can be found sporadically throughout the aged human brain. The prevalence of degenerating microglia increases dramatically in Alzheimer’s disease (AD) and in Down’s syndrome where damaged microglia are colocalized with neurofibrillary degeneration, thus raising the possibility that this kind of neurodegeneration could be a consequence of dwindling microglial neuroprotection. This idea could help resolve the long-standing debate over whether microglial activation is neurotoxic in that one might argue that if overactivation was potentially damaging to neurons, as claimed by numerous authors, a solution for defusing this hazardous situation would be for microglia to die thereby removing the source of danger. Biologically speaking, this would make sense because, unlike most other cells in the CNS, microglia can be replaced and are therefore relatively expendable.

Fig. 2
figure 2

Shown are two examples (a, b) of dystrophic microglial cells from a 91-year-old female AD subject, Braak stage IV (tissue provided by Dr. Thomas Beach, Bannerhealth Research Institute, Sun City, AZ, USA). The ultrastructure reveals widespread vacuolation (vac) throughout the cells’ cytoplasm likely representing dilated ER and Golgi, as well as swollen mitochondria (arrowheads). Nuclei (Nu) are strongly heterochromatic, as is typical for microglia. Nucleus in b is cut in two places. Magnification indicated by scale bars

Full size image
Fig. 3
figure 3

An updated view on microglia functional plasticity involving neuronal-microglial interactions during acute and chronic activation. The initial trigger for microglial activation are signals emitted by damaged or stressed neurons that tell resting cells to divide, increase production of cytokines and growth factors, and change surface antigen expression. This increase in microglial activities, reflected in cellular hypertrophy, represents an acute neuroinflammatory reaction that is designed to bring about neuronal recovery from stress or injury. If successful, excess numbers of activated microglia are eliminated through programmed cell death [88]. If unsuccessful, acute neuronal death will result in the formation of microglia-derived brain macrophages that clear out debris. A chronic neuroinflammatory reaction occurs if damaged neurons continue to send out activating signals that result in persistent microglial activation. Such prolonged overactivation will cause some microglia to become senescent and undergo degenerative changes and eventually lead to widespread microglial degeneration. Once a critical number of microglia have undergone this type of accidental cell death, neurons will have lost all microglial support and are destined to follow a path of slow neurodegeneration, reflected morphologically by abnormal inclusions (e.g. Lewy bodies) and/or neurofibrillary degeneration

Full size image

According to recent research, microglial activation is considered one mechanism by which early-life seizures contribute to increased vulnerability to neurologic insults in adulthood [164]. Thus, microglial cells may develop a functional “memory” causing increased disease susceptibility. However, in a positive sense microglial malleability is likely to represent a key mechanism underlying beneficial functional plasticity of the CNS.

Recent findings also indicate that microglial activation during stress differs qualitatively from those of infection or inflammation [175]. Zinc has been identified as a novel trigger for microglial activation [94], and CD38 (ADP-ribosyl cyclase/cyclic ADP-ribose hydrolase) has been found to link microglial activation to activation-induced cell death by controlling the amount of NO that microglia cells produce [121]. Activation of P2X7 purinoreceptors induces CCL3 production in microglial cells through transcription factor NFAT [92]. In addition, the P2X7R signaling pathway may negatively regulate autophagic flux through the impairment of lysosomal functions, leading to stimulation of a release of autophagolysosomes/phagolysosomes into the extracellular space [177].

Schwartz et al. [159] have made the suggestion that microglia “comprise a family of cells with diverse phenotypes” in order to explain observations on their seemingly contradictory “beneficial” or “destructive” properties under various experimental conditions [159]. We disagree with the concept of ‘bipolar’ microglial activation states and suggest that there is instead a prototypical but highly malleable, i.e. functionally plastic microglial cell type that not only has its unique ontogenesis but which may also become directly affected by disease. Both ontogenetic aspects and its anatomic location could influence a microglial cell’s functional repertoire. The view that microglia can contribute to neuronal diseases is expressed by some authors in the increasingly popular but imprecise term, “non-cell autonomous”.

A family of CNS macrophages consisting of at least five subtypes exists [67]. We believe that their definition and distinction from microglia is of great practical utility as well as conceptual importance. This holds true even if a certain overlap with “bone marrow-derived microglia” exists.

“Inflamed” microglia?

There is probably no area of microglial research that has attracted more attention in recent years than inflammatory phenomena associated with their activation process. The underlying mechanisms, however, have remained largely unclear and the interpretation of some findings related to the “activated” microglial phenotype continues to be problematic. Currently, many authors appear to consider any response of microglial cells to tissue pathology, which is regularly associated with increased expression of MHC class II molecules, as evidence of nervous tissue “inflammation”. The incorrect term, “inflamed microglia” has also been used. However, ‘inflamed’ in pathology refers to interactions of cells and a cellular response to injury, which is a beneficial process unless it becomes chronic or directed against self antigens resulting in autoimmune disease.

The equation ‘detection of MHC class II antigens in microglial cells’ equals ‘evidence of inflammation’ represents a generalization which does not take into account that expression of MHC class II molecules does not suffice to start an immune response, and to prove this point, cells of the peripheral immune system are typically missing in the allegedly “inflamed” brain tissue of AD cortex or Parkinsonian substantia nigra. The fact is that the term “neuroinflammation” is still in need of a consensus definition, and it is unfortunate that it is almost always used in a negative way (i.e. a chronic, destructive process or autoimmune attack). There is clear experimental evidence which shows that activated microglia can be beneficial [49]. In addition, it is increasingly appreciated that interactions between peripheral immune cells and the nervous system differ with each type of insult [73]. The problem will be discussed in detail in the section on neurodegeneration.

The active participation of microglial cells in established autoimmune diseases such as multiple sclerosis is, however, accepted. Interleukin (IL)-17-producing helper T cells may play a pivotal role in the pathogenesis of multiple sclerosis, and IL-17 has been found to upregulate the microglial production of IL-6, macrophage inflammatory protein-2, nitric oxide, adhesion molecules, and neurotrophic factors [95]. Microglia also are the major cellular source of inducible nitric oxide synthase during experimental herpes encephalitis [117]. In bacterial infections, nucleotide-binding oligomerization domain-2 (NOD2), a member of the novel nucleotide-binding domain leucine-rich repeat region containing a family of proteins (NLR) that functions as an intracellular receptor for a minimal motif present in all bacterial peptidoglycans, is functionally expressed by both microglia and astrocytes [27]. Toll-like receptor 4 (TLR4) has a critical role in ethanol-induced microglial activation which is also considered inflammatory [54]. Another recent finding suggests that microglia are able to cross-present exogenous antigens on major histocompatibility complex (MHC) class I molecules to CD8(+) T cells [9]. Importantly, following IFN-gamma stimulation, production of neuroprotective factors by microglia appears to be down-regulated [130].

Role in neurodegeneration

The suggestion that microglial cells pose a threat to neurons in neurodegenerative conditions although presently popular is not backed by firm evidence. The view that microglia are dangerous to nerve cells in vivo is largely based on extrapolation from cell culture studies, non-physiological models such as LPS injection to simulate Parkinson’s disease, or the hypothetical view that any microglial cell that expresses MHC class II molecules and perhaps cytokines is in a state of cytotoxic activation. However, this is not justifiable any longer in view of increasing evidence to the contrary [61, 76, 136, 144]. Furthermore, a large number of distinct brain pathologies exist. The sheer number of different disease states which are considered to be due to unrelated etiologies is incompatible with the simplistic view that for example expression of IL-1 [161] is of the same high pathophysiological importance in each state and across an entire spectrum of very different disorders. It is obvious that a significant number of publications dealing with microglia and inflammation in the field of neurodegeneration, and notably common diseases such as Alzheimer’s or Parkinson’s, have a high propaganda index [89].

Amyotrophic lateral sclerosis

Microglia have been assigned a “non-cell autonomous” function in ALS pathogenesis [18], which is a cryptic way of saying they contribute to motor neuron death. Based on work employing mSOD1-overexpressing BV-2 microglial cell lines it has been suggested that mSOD1 expression in ALS facilitates microglial neurotoxic inflammatory responses via TLR2, which seems mediated by uncontrolled ROS generation [110]. Furthermore, it has been suggested that macrophage colony-stimulating factor (M-CSF) exacerbates ALS in a mouse model through altered responses of microglia expressing mutant superoxide dismutase [62]. ALS microglia have been implicated as a source of the increased monocyte chemoattractant protein-1 (MCP-1) levels which are detected in ALS patients and in the ALS mouse model [155]. A review of the proposal that ROS mediate ALS pathogens demonstrates that a new twist has been introduced: it is ROS produced by microglia [17]. However, Gowing and co-workers have found that ablation of proliferating microglia does not affect motor neuron degeneration in amyotrophic lateral sclerosis caused by mutant superoxide dismutase, allowing these authors to conclude that “microglia are fundamentally neuroprotective cells but that expression of mutant SOD1 renders them neurotoxic” [61]. In line with this conclusion is an alternative view of microglial involvement in ALS expressed by Fendrick et al. [53], namely, that motoneuron death is the result of microglial malfunction and, ultimately, degeneration occurring during the course of the disease, thus attributing neurodegeneration to a loss of microglial neuroprotection.

Alzheimer’s disease

One of us [171] has demonstrated that microglial dystrophy coincides spatially with presence of and precedes temporally the spread of tau pathology in AD. Deposits of amyloid-beta protein (Abeta) devoid of tau-positive structures were found to be colocalized with non-activated, ramified microglia, suggesting that Abeta does not trigger microglial activation. The findings further indicate that when microglial activation does occur in the absence of an identifiable acute central nervous system insult, it is likely to be the result of systemic infectious disease. These findings argue strongly against the hypothesis that neuroinflammatory changes contribute to AD dementia [171] but do not exclude interactions between microglia and beta-amyloid deposits.

Microglia have indeed been shown to play a role in amyloid removal, particularly compacted amyloid deposits, under certain conditions [131]. Microglia also internalize soluble Abeta from the extracellular milieu through a nonsaturable, fluid phase macropinocytic mechanism that is distinct from phagocytosis and receptor-mediated endocytosis both in vitro and in vivo [115]. Abeta activates a purinergic autocrine/paracrine stimulatory loop in microglia of which the P2X(7) receptor is an obligate component [154]. The TLR2 signaling pathway has been implicated in the mediation of fibrillar Abeta peptide-induced activation of microglial cells [85], and microglia activation has been suggested to mediate fibrillar amyloid-beta toxicity in the aged primate cortex [103].

Alterations in levels of signaling molecules secreted by microglia may in part be responsible for the defective neurogenesis observed with FAD-linked mutant PS1 in vivo, a finding supporting a non-cell-autonomous role for mutant PS1 in hippocampal neurogenesis [32]. Indeed, microglia expressing FAD-linked PS1 variants impair proliferation of neural progenitor cells [32]. This supports our view that ‘sick glia’ in degenerative CNS diseases may have a direct pathogenic role [35].

It has become common practice in neurobiology to extrapolate from mouse models to human pathology. However, there are great differences even between rodent species as exemplified by the propensity of mouse CNS to become infiltrated by peripheral inflammatory cells. For instance, following peripheral nerve axotomy mice show significant infiltration of CNS tissue by peripheral immune cells whereas rats do not [146]. Therefore, experiments utilizing inflammatory infiltration in the CNS of AD mice [50, 51] have to be interpreted with caution. In fact, recent experimental evidence argues strongly against inflammation as a driving force for amyloid deposition [26]. Furthermore, at least in mice, neither amyloid plaque formation and maintenance nor amyloid-associated neuritic dystrophy depends on the presence of microglia [71].

Parkinson’s disease

The involvement of microglia in the pathogenesis of Parkinson’s disease (PD) remains controversial. We do not think that the currently available evidence provides sufficient support for the hypothesis that microglial cells, or “microglial inflammation”, exert an active, disease-promoting role in PD. This critical view is based on a number of facts. First, a mechanism by which activated microglia might specifically target dopaminergic neurons is missing [151]. Second, T cells are not usually observed in Parkinsonian substantia nigra and a report claiming the presence of small numbers of scattered cells [21] is awaiting independent confirmation. Thus, the title of a recent publication, “T cell-microglial dialogue in Parkinson’s disease and amyotrophic lateral sclerosis: are we listening?” [3] is likely to cause confusion. Third, a popular LPS injection model of PD [80] which has been used to provide experimental support for the role of ‘microglial inflammation’ in PD seems quite artificial. Fourth, an infectious etiology [86] does not appear to be of relevance in the vast majority of PD cases because the underlying concept does not do justice to other, well-established findings in PD, such as deficits in mitochondrial function, the role of oxidative and nitrosative stress, dysfunction of the ubiquitin–proteasome system (UPS) and functional defects caused by disease-causing mutations whose role in PD is well established [64]. Finally, it is worth mentioning that in contrast to a study in mice [100], the human substantia nigra does not contain the highest density of microglial cells [124].

Speculations on an active role of microglial cells in PD were originally based on their expression of MHC class II molecules in the Parkinsonian nigra [122]. However, MHC class II antigen expression by microglia in the substantia nigra cannot be used as an indicator of clinical PD severity or disease progression [36]. We have previously suggested that microglial MHC class II expression that is widespread in neurodegenerative diseases could serve to down-regulate rather than up-regulate antigen presentation capacity and thus serve as a protective “firewall” against unwanted immune attacks when it occurs in the absence of co-stimulators [63]. This hypothesis is in line with the finding of CD163 upregulation by nigral microglia specifically in PD cases with a shorter disease course. Because CD163 has been implicated in anti-inflammatory signaling pathways, this observation suggests that “microglial inflammation” is modulated, rather than self-propagating, in PD [37]. In addition, from a systems biological point of view, only a smaller part of the emerging common pathway of PD [128] shows links to immunological or inflammatory mechanisms. The latter might include Nurr1 modulation in microglia and astrocytes [52, 153]. Overall, the biological association between PD and “inflammatory” parameters may have to be interpreted in a modified way which takes the specifics of nervous tissue into account.

In keeping with the view that microglia in PD are reactive rather than aggressive, synuclein has been found to stimulate microglia [174]. It has also been observed that the homocysteine-induced endoplasmic reticulum protein (herp) is up-regulated in PD substantia nigra [163]. Relevance of herp for PD is also supported by another study [31]. Homocysteine promotes proliferation and activation of microglia [204], and it has been proposed that it may be promising to regulate microglia activation in PD via targeting CD200–CD200R signaling pathways [192].

Taken together, we find the hypothesis attractive that similar to AD [166] and Huntington’s disease [162], glia in PD may be compromised [48]. In other words, microglia themselves may be affected by the disease process, and may therefore not be capable of exerting sufficient neuroprotective functions such as glutathione peroxidase expression [144]. In line with this view, chronic nicotine treatment increases the number of HLA− DR+ microglia in monkey substantia nigra after nigrostriatal damage and results in improved nigral integrity and function [145].

Macular degeneration

Retinal microglia have also been implicated in disease [30]. For instance, CX3C chemokine receptor 1 (CX3CR1)-dependent subretinal microglia cell accumulation is associated with cardinal features of age-related macular degeneration (AMD) [34], and the cells have been suggested to contribute to the progression of this most common cause of blindness in the elderly [29]. Amyloid-beta which is a major component of retinal drusen (plaques) in AMD has been found to up-regulate complement factor B in retinal pigment epithelial cells through cytokines released from recruited macrophages/microglia [190]. However, whether microglia activated by primary rod cell death actively kill adjacent photoreceptors [72] remains to be established. Finally, it should be noted that normal aging also takes its toll on microglia [167, 172].

Microglia in brain tumors

This is a rapidly developing field. There can be little doubt that the macrophage population in brain tumors is recruited from more than one source, i.e. brain and blood, although the mechanisms of macrophage tumor-attraction and invasion are unclear. The hypoxia-inducible factor (HIF)-1alpha-CXCR4 pathway is of particular interest because it has been implicated in microglia migration [191]. Interestingly, HIF-1alpha also blocks differentiation of malignant gliomas [112]. Tumor cell-derived MCP-3, but not MCP-1 [43], facilitates the infiltration of macrophage/microglia into tumor tissues [139]. In contrast, a role for the chemokine system CX3CL1 and its receptor CX3CR1 has been excluded at least in the GL261 murine model of malignant glioma [107]. Historically, a tumor cytotoxic role of microglial cells has been proposed and apoptosis of glioma cells can be induced by microglia-secreted molecules, such as nitric oxide and cathepsin B [81]. However, under normal conditions microglia support glioma growth. We first made this surprising observation more than a decade ago [70], and the effect has been confirmed by other authors [39, 40]. More recent discoveries in this area include the induction of microglial MT1-MMP by glioma cells [116], and the regulation of glioblastoma invasion by microglia-derived TGF-beta [193]. In contrast, the relevance of some immunological phenomena in high-grade gliomas remains unclear. Compared with human control brains, a significant increase in the percentage of parenchymal IL-16+ macrophages/microglia was observed in grade II astrocytomas, a further increase was observed at the transition from grade II to III astrocytomas, and this increase in IL-16 immunoreactivity correlated with WHO grades of human astrocytic brain tumors [105].

Gemistocytic astrocytomas contain unusually high numbers of microglial cells, and the higher the number of class II immunoreactive gemistocytes, the fewer class II positive microglial cells can be detected. This may be related to the especially poor prognosis of gemistocytic astrocytomas for which induction of T cell anergy could provide one explanation [58].

CNS trauma

An understanding of the molecular mechanisms underlying microglial “activation”, and clarification of where and when alternative macrophage activation pathways [119] are involved, will also be of importance for our understanding of the role microglia play in CNS trauma. It has been suggested that microglia mediate secondary tissue damage in traumatic brain injury but this does not represent a well-founded speculation. Early infiltration of lesions of traumatic brain injury in rats by CD8+ macrophages/microglia has been reported, and the cells co-expressed endothelial monocyte-activating polypeptide II and P2X4 receptor indicating a central role for these factors in lesion development [200]. Lünemann and co-workers who believe that microglia cause secondary damage after brain trauma [113] have characterized a new macrophage/microglia activation factor, MAF and propose that MAF expression plays a functional role in the differentiation of microglia into a phagocytosing phenotype and that MAF may be required for phagocytotic activity, specifically in traumatic tissue lesions. Brain trauma induces expression of diacylglycerol kinase zeta in microglia [134].

Transection of the adult rat facial nerve leads to an increase in the number of microglia in its central nucleus of origin, with a peak of proliferation 3–5 days after axotomy. These proliferating microglia do not produce deleterious factors for neurons but express macrophage-colony stimulating factor (M-CSF) and granulocyte macrophage-CSF (GM-CSF) as well as their specific receptor proteins, c-Fms and GM-CSFRalpha [133]. Microglia also promote astrogliogenesis and maintenance of neural stem cells by activating Stat3 function and via notch and sox9 signaling pathways [203].

Microglia and neuropathic pain

Recent findings challenge the neuron-centric view of neuropathic pain etiology and pathology [198]. There is a rapidly growing body of evidence indicating that microglial cells have a causal role in the pathogenesis of pain hypersensitivity following nerve injury [83]. Neuropathic pain, a debilitating pain condition that commonly occurs after nerve damage, is a reflection of the aberrant excitability of dorsal horn neurons [183]. The chemokine CCL2, produced by both damaged and undamaged primary sensory neurons, is a key mediator of microglia activation in neuropathic pain states [179]. Its receptor, CCR2 appears to be critical in the activation of spinal microglia and is expressed in both resident and bone marrow-derived cells [198]. The receptor for the cytokine IFN-gamma, IFN-gammaR also appears to be a key element in the molecular machinery through which resting spinal microglia transform into an activated state that drives neuropathic pain [183]. Furthermore, ligation of L5-L6 spinal nerves evokes an accumulation of active, phosphorylated STAT3, mediated mainly via IL-6 signaling, in microglial cells of dorsal spinal cord mostly in projection areas of injured nerves [46]. Microglia have been shown to be involved in intestinal pain [20]. Both astrocytes and microglia of the locus coeruleus are involved in cardiac pain processing after acute cardiac injury [199].

Cannabinoid receptor type 2 (CBR2) inhibits microglial reactivity and mitogen-activated protein kinase-phosphatase (MKP)-3 induction is part of the mechanism underlying the reduction of pro-inflammatory factors and microglial migration caused by CBR2 agonists [152]. Interleukin-18-mediated microglia/astrocyte interactions in the spinal cord enhance neuropathic pain processing after nerve injury [125]. In contrast, the anti-inflammatory cytokine interleukin-10 reduces the number of spontaneous flinches in the early and delayed phases of the formalin test of inflammatory pain, and this correlates with a block in phosphorylation of p38 and reduced expression of 26 kDa full-length membrane spanning tumor necrosis factor-alpha in spinal microglia [202]. It has been hypothesized that microglial TNFalpha may enhance the efficacy of synaptic transmission [11].

Extracellular nucleotides, which act through purinoceptors, especially the G-protein-coupled P2Y receptor P2Y(12)R, have been implicated as signaling molecules used by microglia to sense adverse physiological conditions such as neuronal damage [181]. P2Y(12)R is a metabotropic ATP receptor that is thought to be involved in the pathogenesis of neuropathic pain [96, 181]. P2X(4) receptors are expressed de novo by activated microglia in the spinal cord in response to nerve injury and mediate BDNF release as well as neuropathic pain [184]. A hypothesis has been formulated that P2X(7) receptors on the presynaptic terminal and those on the microglia synergistically act to ensure feedback pathways that reset to a high level the efficacy of synaptic transmission, thus ensuring chronic neuropathic/neuroinflammatory pain even when the initial insult has subsided [11].

Loss of glycinergic and GABAergic synaptic inhibition (i.e. dis-inhibition) in the spinal cord dorsal horn has been increasingly recognized as an important process in the development and maintenance of chronic pain of both inflammatory and neuropathic origin [197]. The production of inflammatory mediators and/or the activation of immune cells appear to be critical for both and a model of microglia-induced dis-inhibition in neuropathic pain states has been proposed [197]. Remote activation of microglia and pro-inflammatory cytokines predict the onset and severity of below-level neuropathic pain after spinal cord injury in rats [44].

Future perspectives

Microglia imaging—the future has begun

Microglia research entered a new era when it became possible to study microglia activation in vivo [6, 7, 24]. In the context of microglia imaging, mitochondria have become a subject of special interest [8]. The spectrum of suitable positron emission tomography ligands has been extended in the meantime [114, 185, 186], and it has also become possible to watch microglial cells patrolling and at work in the living brain using two-photon microscopy [137]. In vivo imaging of microglial cells is now also being performed in fish in order to understand basic neurobiological mechanisms [142]. However, it should be pointed out that there are known species differences even between mice and humans concerning macrophage activation pathways [119]. In addition to ex vivo live imaging [101], biophotonic/bioluminescence molecular imaging has been used recently to visualize transcriptional activation of microglial cells in the brains of live animals [99]. Attempts to target activated microglia in a transgenic mouse model of Alzheimer’s pathology by means of intraventricular delivery of a phagocytosable MRI contrast agent have been made [132]. Microglia containing gadolinium-grafted nanoparticles have also been employed for in vivo imaging [148].

Bone marrow-derived microglia

There has been significant confusion in the literature over many years as to whether and if so to what extent microglia and macrophage precursors derived from the bone marrow exchange in the adult brain. Recent studies appear to settle the issue and it has been suggested that certain pathological conditions can trigger the recruitment of new, bone marrow-derived microglia [55]. However, microglia progenitor recruitment from the circulation in denervation or CNS neurodegenerative disease conditions remains a controversial issue [1, 123]. If it occurs at all in non-experimental conditions, it should be extremely limited. However, the ontogenetic link between bone marrow-derived cells and microglia now seems firmly established [41, 59, 165]. Interestingly, the functional phenotype of bone marrow-derived microglia appears to differ from that of resident cells [91]. Microglia recruit monocytes into the brain during peripheral organ inflammation [38]. Conversely, microglia- and bone marrow-derived dendritic cells can leave the CNS via the blood stream [77].

Microglia cell-based therapies

The use of genetically modified microglia precursors to study microglial cells in adult CNS has long been envisioned [65]. The development of microglia-based cell therapies now appears to be within reach [41, 68, 135]. A number of therapeutic uses of microglia have been proposed.

Hematoma resolution may be one potential therapeutic application [201]. Microglia have also been proposed as vehicles for inducible thymidine kinase gene therapy [148]. Nanotubes which are taken up by microglia could be utilized as novel vehicles for targeted therapy in brain cancers [93] although precautions have to be taken because the side-effects of nanotubes have not been sufficiently studied. While depletion of peripheral macrophages and brain microglia increases therapeutically desirable brain tumor titers of oncolytic viruses [56], irradiation-induced loss of microglia occurs in the young brain possibly contributing to the development of cognitive deficits [90].

A number of attempts have been made to control microglial activation through pharmacological intervention ranging from herbal medicine [109] to antibiotics [74] and neurotransmitter analogs [195]. Activation of adenosine 2A receptors after intrathecal administration may be a novel, therapeutic approach for the treatment of neuropathic pain by increasing IL-10 in the immunocompetent cells of the CNS [111]. IL-10 production is reduced by hypothermia but augmented by hyperthermia in rat microglia [120]. Modulation of microglia is of special interest for the potential treatment of inflammatory disease conditions. Moreover, enhanced microglial activation and proinflammatory cytokine upregulation have been linked to increased susceptibility to seizures and neurologic injury [164].

It appears that microglia can be used to both minimize cytotoxicity and maximize neuroprotection in neurodegenerative diseases. Specifically, wild-type microglia extend survival in PU.1 knockout mice with a familial amyotrophic lateral sclerosis mutation [10]. Similarly, activation of bone marrow-derived microglia promotes photoreceptor survival in inherited retinal degeneration [156]. Resident microglia and recently bone marrow-derived cells are assumed to co-operate [87]. Furthermore, bone marrow-derived mesenchymal stem cells reduce brain amyloid-beta deposition and accelerate the activation of microglia in an acutely induced Alzheimer’s disease mouse model [102]. Finally, microglia have been recently implicated in volume recovery in non-neurogenic brain regions during abstinence after alcohol dependence preceding neurogenesis [138].