Neurobiology of microglial action in CNS injuries: receptor-mediated signaling mechanisms and functional roles

Abstract

Microglia are the first line of immune defense against central nervous system (CNS) injuries and disorders. These highly plastic cells play dualistic roles in neuronal injury and recovery and are known for their ability to assume diverse phenotypes. A broad range of surface receptors are expressed on microglia and mediate microglial ‘On’ or ‘Off’ responses to signals from other host cells as well as invading microorganisms. The integrated actions of these receptors result in tightly regulated biological functions, including cell mobility, phagocytosis, the induction of acquired immunity, and trophic factor/inflammatory mediator release. Over the last few years, significant advances have been made towards deciphering the signaling mechanisms related to these receptors and their specific cellular functions. In this review, we describe the current state of knowledge of the surface receptors involved in microglial activation, with an emphasis on their engagement of distinct functional programs and their roles in CNS injuries. It will become evident from this review that microglial homeostasis is carefully maintained by multiple counterbalanced strategies, including, but not limited to, ‘On’ and ‘Off’ receptor signaling. Specific regulation of theses microglial receptors may be a promising therapeutic strategy against CNS injuries.

1. Introduction

Microglia were first recognized as a distinct cellular entity in the central nervous system (CNS) by the German anatomist Franz Nissl and subsequently given their name by the Spanish neuroscientist Pío del Río-Hortega between 1919 and 1921. Over the course of the past century, much evidence has accumulated on the importance of this cell population in CNS homeostasis and its involvement in CNS pathologies. Similar to the role of peripheral macrophages, microglia are now known as the first line of defense against CNS injuries, including stroke, traumatic brain injury, and spinal cord injury. Following an insult, resident microglia rapidly mobilize to the injury site, where they play a role in acute damage and modulate the long-term progression of injury. Whether this microglial activation in the compromised CNS is helpful or destructive remains controversial. In support of a beneficial role of microglia, selective depletion of proliferative microglia is known to exacerbate ischemic brain injuries (Faustino et al., 2011; Lalancette-Hebert et al., 2007), whereas injection of microglia into the ischemic brain is known to ameliorate injuries (Imai et al., 2007). Microglial activation is thought to benefit the injured brain by removing cell debris and restoring tissue integrity (Hanisch and Kettenmann, 2007; Kwon et al., 2013; Miron et al., 2013; Thored et al., 2009). In contrast, mounting evidence reveals that inappropriate or excessive microglial activation may lead to secondary expansion of brain damage and deterioration of neurological outcomes (Barone and Feuerstein, 1999; Dirnagl et al., 1999). The toxicity of microglia is mediated by the release of a plethora of harmful substances, including nitric oxide (NO), reactive oxygen species (ROS), and proinflammatory cytokines (Block et al., 2007). Microglia can also impair neurogenesis (Ekdahl et al., 2003; Liu et al., 2007), oligodendrogenesis (Miron 2013) and prevent axon regeneration (Schwab and Bartholdi, 1996). As a result of findings such as these, it is increasingly well-accepted that microglia play dualistic roles in neural injury and recovery and can improve or destroy tissue integrity depending upon the cellular context. An improved understanding of the mechanisms underlying microglial activation and their functional modulation by the local microenvironment is likely to advance our knowledge of many CNS pathological states.

Because of the dual-faced nature of microglia in the CNS, their activity must be tightly regulated so that they can be promptly turned on as the first responders to noxious stimuli and then rapidly turned off to avoid unwanted side effects. In general, microglia communicate with other CNS cells in one of two ways. First, microglia are able to recognize the signals secreted and released by other CNS cells from a considerable distance. Second, microglia bind to surface molecules expressed on adjacent CNS cells. These microglia-regulating signals can be further divided into two broad categories, so-called ‘Off’ and ‘On’ signals (Biber et al., 2007). The ‘Off’ signals are usually constitutively expressed or released under physiological, resting conditions. Under pathological conditions, either the loss of the ‘Off’ signal or the gain of the ‘On’ signal initiates the rapid activation and mobilization of microglia. Microglia sense these signals through a wide array of recognition receptors. Specific regulation of these receptors with pharmacological and other tools thus represents a promising therapeutic strategy against CNS injuries.

In this review, we will describe the current state of knowledge on microglial surface receptors and their roles in CNS injuries, with a special emphasis on their engagement of distinct functional programs. This information leads to an improved understanding of how different receptors work in concert to maintain microglia in a state of balanced equilibrium. It will become evident from this review that microglial homeostasis is carefully maintained by multiple opposing and complementary switches, including the ‘On’ and ‘Off’ receptors and their downstream signaling events. These opposing signals appear to have evolved such that they are normally activated only when needed at the site of injury and subsequently inactivated to prevent excessive release of toxic mediators. Indeed, the counterbalancing of these opposing signals is a hallmark of a well-balanced and healthy CNS immune system (Figure 1). When this homeostatic equilibrium is disrupted, neurological injury may be exacerbated. Thus, the carefully timed modulation of microglial receptor activity is a rational therapeutic strategy.

Figure 1. Microglial receptors.

Various surface receptors act as opposing and complementary switches to maintain microglial homeostasis. “On” receptors: A large number of receptors have been identified to turn on microglial activation in response to different stimuli. “Off” receptors: Several receptors constitutively expressed on resting microglia help to maintain their quiescence. Dualistic receptors: The activation of these receptors induces both activating and inhibitory effects in microglia, allowing for a high degree of resolution in microglial responsiveness to differing stimuli. P1: P1 adenosine receptors; P2: P2 ATP receptors; RAGE: receptor for advanced glycation endproducts; TLR: toll-like receptor; TREM: triggering receptors expressed on myeloid cells.

2. Microglial receptors in the immunoglobulin superfamily

The receptors in the immunoglobulin superfamily (IgSF) consist of proteins with one or more IgSF structural domains. Recently, this family of receptors has been the focus of much attention because several of its members recognize the ‘Off’ signal and maintain microglia in a normal, surveillent status. In general, these ‘Off’ receptors signal through a cytoplasmic-domain immunoreceptor tyrosine-based inhibition motif (ITIM). In contrast, the ‘On’ receptors in this family sense pathological changes in neurons and other CNS components and signal through a cytoplasmic-domain immunoreceptor tyrosine-based activation motif (ITAM), thereby leading to microglial activation (Linnartz and Neumann, 2013). These opposing roles of ‘Off’ versus ‘On’ receptors and ITIM versus ITAM sigaling are typical of the counterbalancing strategies that determine the direction of microglial responses to fluctuations in the environmental milieu and maintain microglial equilibrium. Therefore, we will discuss several receptors in this family in the following section.

2.1. Triggering receptors expressed on myeloid cells (TREM)

TREM is a family of cell surface receptors specifically expressed on myeloid cells. The trem gene cluster is located on human chromosome 6p21 and mouse chromosome 17C3, and encodes TREM-1, TREM-2, TREM-3 (in the mouse), and other TREM-like genes (Klesney-Tait et al., 2006). The activation of TREM-1 enhances inflammatory responses while TREM-2 activation inhibits inflammation (Klesney-Tait et al., 2006), again exemplifying the principle of counterbalanced activating and inhibitory immune responses. TREM-2 is most relevant to the CNS by virtue of its constitutive expression on microglia and its proven role in microglia-neuron interactions.

2.1.1. TREM-2 mediates microglial clearance activity

TREM-2 has been detected in microglia both in vitro and in vivo by in situ hybridization and immunohistochemical staining (Kiialainen et al., 2005; Schmid et al., 2002; Sessa et al., 2004; Thrash et al., 2009). Microglial expression of TREM-2 is highly heterogeneous across brain regions, with high levels of TREM-2-expressing microglia in the lateral entorhinal cortex, cingulate cortex, and white matter, whereas few, if any, TREM-2-expressing microglia are found in the hypothalamus, circumventricular organs, or the median eminence. Even in areas with a relatively high proportion of TREM-2-expressing microglia, TREM-2-positive and TREM-2-negative microglia tend to be intermixed. These differences in TREM-2 expression in the healthy brain may reflect the diversity in microglial responsiveness and may contribute to regional variations in sensitivity to pathological signals.

The intracellular distribution of TREM-2 within microglia also shows distinct features. The majority of TREM-2 is located intracellularly in the Golgi complex and in exocytotic vesicles (Prada et al., 2006; Sessa et al., 2004). This intracellular pool of TREM-2 is dislodged and migrates to the cell surface following ionomycin or interferon-gamma (IFN-γ) exposure, resulting in transient elevation of surface TREM-2 (Sessa et al., 2004). Another study reported that microglial TREM-2 mRNA expression is rapidly decreased in response to lipopolysaccharide (LPS) or LPS+IFN-γ stimulation (Schmid et al., 2002). All these studies suggest that microglial TREM-2 responds rapidly to inflammatory signals by alterations in transcription and subcellular distribution, supporting the notion that TREM-2 is exquisitely sensitive and that it may regulate microglial activity. Indeed, TREM-2 overexpression and knockdown studies in microglia have demonstrated that TREM-2 stimulation promotes microglial migration and the phagocytosis of neuronal cell debris without inducing inflammatory cytokine expression (Takahashi et al., 2005). Such special ‘turning on’ properties render TREM-2 a reasonable candidate for directing the benefical activation of microglia.

The first exploration of the role of TREM-2 in CNS pathologies was initiated upon the genetic analysis of a recessively inherited disorder, polycystic lipomembranous osteodysplasia with sclerosing leukoencephalopathy (PLOSL, otherwise named Nasu-Hakola disease). This disease is characterized by insidious and progressive presenile dementia associated with multiple bone cysts. Disruption of TREM-2 or its adaptor protein DNAX-activating protein of molecular mass 12 kilodaltons (DAP12) was identified as the causal mutation underlying PLOSL (Paloneva et al., 2002). TREM-2 deficiency may result in insufficient clearance of apoptotic cells and exaggerated inflammation, which may elicit the frank CNS degeneration that is evident in PLOSL patients. Interestingly, the loss of TREM-2 seems to be important for white matter pathology because PLOSL patients exhibit sclerotic lesions that include activated microglia in the white matter and myelin pallor (Verloes et al., 1997). Consistent with this notion are the frequent close appositions of TREM-2-positive microglia with oligodendrocytes during development (Thrash et al., 2009) and the predominant distribution of TREM-2-positive microglia in the white matter of the adult brain (Schmid et al., 2002).

Recently, microglial TREM-2 expression has also been shown to be involved in other neurological disorders such as multiple sclerosis (MS) (Piccio et al., 2007) and Alzheimer’s disease (AD) (Frank et al., 2008). Some studies also reveal increased microglial expression of TREM-2 after ischemic stroke (Heldmann et al., 2011; Sugimoto et al., 2014). The function of TREM-2 in acute brain injuries, however, has not yet been defined.

2.1.2. TREM signaling

Despite an awareness of the importance of TREM-2 in microglial function, the ligands of TREM-2 have not yet been identified. This has hindered our understanding of microglial activation through this receptor. Using TREM-Fc fusion proteins, Hsieh and colleagues have demonstrated that multiple types of cultured neuronal cells bind to TREM2-Fc but not TREM1-Fc, suggesting that TREM-2 ligand(s) are constitutively expressed on healthy neurons (Hsieh et al., 2009). In addition, apoptotic signals were found to enhance the neuronal expression of TREM-2 ligand(s). The Hsieh study further demonstrated that TREM-2 and TREM-2 ligand(s) form a receptor-ligand pair connecting microglia with apoptotic neurons and directing the phagocytic removal of damaged cells. However, the specific identity of TREM-2 ligand(s) remains elusive. Another study has shown that heat shock protein (HSP) 60, a mitochondrial chaperone that can also be expressed at the cell surface, is a TREM-2-binding protein exposed on the surface of neurons as well as astrocytes (Stefano et al., 2009). Accordingly, treatment with HSP60 was found to stimulate microglial phagocytosis in a TREM-2-dependent manner.

In contrast to the idea of a single specific ligand for TREM-2, Takegahara and colleagues have documented that TREM-2 associates with another receptor on the cell surface of dendritic cells named Plexin-A1, and forms a higher order receptor complex containing TREM-2, plexin-A1, and DAP12 (Takegahara et al., 2006). The activation of plexin-A1 by its ligand, semaphorin 6D, can be blocked by decreasing either TREM-2 or DAP12 expression, suggesting a functional connection between plexin-A1/semaphorin 6D and TREM-2/DAP12 signaling. In short, the exploration of TREM-2 ligands may involve more than the identification of a single binding partner and may require an approach encompassing possible co-receptors. Furthermore, it still remains to be determined whether similar receptor complexes exist in microglia that regulate TREM-2–mediated communication of microglia with other CNS cells.

TREM-2 and TREM-1 both have very short cytoplasmic domains and lack intrinsic signaling capacity (Bouchon et al., 2000). Thus, both must signal through the transmembrane adaptor molecule DAP12. DAP12 contains an ITAM that can bind to the positively charged lysine residue in the transmembrane domain of TREM. Upon the engagement of TREM receptors, Src kinases phosphorylate the tyrosine residues in the DAP12 ITAM, forming a docking site for the binding of protein tyrosine kinase Syk (Ivashkiv, 2009; N’Diaye et al., 2009). Despite the similarity between TREM-2 and TREM-1 signaling, distinct downstream signaling pathways are activated by these two receptors beyond this point in order to maintain microglial equilibrium. Upon TREM-1 ligation in inflammatory phagocytes, Syk activates signaling cascades involving ERK1/2, PLCγ, the NFκB transcription factor, and MAPKs (Bouchon et al., 2000; Gibot et al., 2004), thereby leading to cell activation. In contrast to the putative activating effects of TREM-1/DAP12 signaling, TREM-2/DAP12 signaling results in diverse cellular responses ranging from increased cell activation to propagation of inhibitory signals within the cell. In microglia, stimulation of TREM-2 activates ERK1/2, leading to reorganization of the actin cytoskeleton and enhanced phagocytosis (Takahashi et al., 2005). TREM-2/DAP12 signaling may also stimulate the expression of chemokine receptors and promote microglial migration toward chemokine ligands CCL19 and CCL21. Interestingly, TREM-2 also induces inhibitory signals that ameliorate microglial inflammatory responses during the phagocytosis of apoptotic neurons (Takahashi et al., 2005).

TREM-2 can also relay ‘paradoxical’ signaling (activating and inhibitory) within microglia, although the mechanisms are unclear. Nonetheless, the ITAM-containing adapter DAP12 seems to be a critical player (Barrow and Trowsdale, 2006; Peng et al., 2010). A consensus ITIM sequence (SPYQEL) is embedded in the ITAM sequence of DAP12 (ESPYQELQGQRSDVYSDL) and is known as a ‘closet’ ITIM. This closet ITIM may recruit inhibitory phosphatases [eg. SH2 domain-containing inositol polyphosphate 5′ phosphatase (SHIP) and SH2 domain-containing phosphatase-1 (SHP1)] in order to attenuate DAP12 activation (Barrow and Trowsdale, 2006). This hypothesis has already found support in studies of macrophages, where SHIP1 is recruited to DAP12 in response to the ligation of TREM-2 and inhibits TREM-2/DAP12 signaling (Peng et al., 2010). The interaction between SHIP1 and DAP12 requires the binding of the SHIP1 SH2 domain with the phosphorylated tyrosine in the DAP12 closet ITIM. In the absence of SHIP1, ligation of TREM-2 recruits PI3K p85 subunit to DAP12 ITAM with the assistance of another adaptor protein, DAP10 (Peng et al., 2010). This, in turn, results in the activation of ERK1/2, Akt, and the guanine nucleotide exchange factor Vav3 (Figure 2). The factors that dictate the differential response of DAP12 to TREM-2 ligation remain to be elucidated. Barrow and colleagues have proposed that DAP12 might be either incompletely or completely phosphorylated depending on the binding affinity between the associated receptor and its ligands. Incomplete phosphorylation of DAP12 upon weak TREM-2 receptor binding may result in ITIM phosphorylation and lead to recruitment of inhibitory phosphatases. In contrast, high affinity receptor binding may result in complete ITAM phosphorylation and sustained activation (Figure 2). TREM-2 ligands will have to be identified to validate this hypothesis. An alternative possibility is that closet ITIM may extinguish ITAM signaling after the desired activation threshold is achieved, in order to prevent excessive activation.

Figure 2. TREM-2 signaling.

The ITAM-containing adapter DAP12 is important for both activating and inhibitory TREM-2 signaling. A consensus ITIM sequence is embedded in the ITAM sequence of DAP12 and is called a ‘closet’ ITIM. DAP12 might be incompletely or completely phosphorylated depending on the binding affinity between the associated receptor and its ligands. While incomplete phosphorylation of DAP12 upon weak TREM-2 receptor binding can result in ITIM phosphorylation, high affinity receptor binding may result in complete ITAM phosphorylation. The phosphorylated tyrosine in the DAP12 closet ITIM may recruit inhibitory phosphatases (SHP-1 or SHIP1) to attenuate TREM-2 activation. In the absence of SHIP1, ligation of TREM-2 recruits PI3K p85 subunit to the DAP12 ITAM with the assistance of another adaptor protein, DAP10. This, in turn, results in the activation of ERK1/2, Akt,, and the guanine nucleotide exchange factor Vav3.

The evident complexity of TREM signaling, as detailed above, may allow for a high degree of resolution in microglial responsiveness to differing stimuli and remains a fruitful and active area of investigation. However, there is still a dearth of information on the involvement of microglial TREM signaling in pathological states.

2.2. Fcγ Receptors (FcγRs)

FcγRs bind specifically to the Fc (fragment, crystallizable) region of IgG immunoglobulin antibodies and are therefore a gateway to responsiveness to various immune stimuli. As expected, FcγRs are involved in a variety of microglial biological functions that include phagocytosis, cytokine production, and oxidative bursts (Song et al., 2002; Ulvestad et al., 1994; Vedeler et al., 1994).

2.2.1. FcγRs differentially regulate microglial function

There are four principal types of FcγR: FcγRI (CD64), FcγRII (CD32), FcγRIII (CD16) and FcγRIV, which bind different isotypes of IgG with varying affinities (Nimmerjahn and Ravetch, 2008). FcγRs can induce both activating and inhibitory responses in microglia, as one might expect from the overarching need to maintain microglial homeostasis. For instance, FcγRI, FcγRIII and FcγRIV are activating receptors. In contrast, FcγRII (FcγRII in mice and FcγRIIb in humans) is an inhibitory feedback receptor that dampens responses triggered by activating receptors (Syam et al., 2010). Cultured human microglia have been shown to express mRNA for both activating (FcγRI and FcγRIII) and inhibitory (FcγRIIb) FcγR receptors (Lue and Walker, 2002). FcγR stimulation may regulate microglial phagocytosis of IgG-opsonized particles and enhance superoxide and nitric oxide production (Bard et al., 2000; Le et al., 2001; Ueyama et al., 2004). FcγRs are also important for microglial secretion of multiple chemokines, including macrophage inflammatory protein 1α (MIP-1α), monocyte chemotactic protein-1 β (MCP1β), and regulated on activation, normal T cell expressed and secreted (RANTES) in response to immobilized IgG antibodies (Lendvai et al., 2000). Studies using specific IgG antibody-coated Cryptococcus neoformans immune complexes have shown that FcγRI and FcγRIII are critical for MIP-1α production in microglia. Blockade or genetic deficiency of FcγRI and FcγRIII, but not FcγRII, significantly reduces microglial MIP-1α induction (Song et al., 2002). The function of inhibitory FcγRIIb in microglia still awaits identification.

The importance of FcγRs in the pathology of cerebral ischemia was shown by Komie-Kobayashi and colleagues in a mouse model of transient middle cerebral artery occlusion (MCAO, Komie-kobayashi, 2004). FcγR-deficient mice were protected from the progression and expansion of infarct volume after MCAO. This was accompanied by reductions in microglial activation, macrophage infiltration, and microglia/macrophage iNOS production. These findings are consistent with the abovementioned role of FcγRs in the secretion of multiple chemokines and the enhancement of superoxide and NO production. Another seminal study showed that intravenous infusions of immunoglobulin (IVIG), which contain pooled IgG immunoglobulins extracted from the plasma of multiple blood donors, protected against experimental stroke (Arumugam et al., 2007). Although such protection has been attributed to a direct effect of IVIG on neurons, the possibility that IVIG may also work through FcγRs on microglia as well as other immune cells cannot be excluded. Although the Arumugam study appears to contradict the Komie-Kobayashi study, one might speculate that IVIG preferentially engaged the inhibitory microglial FcγRs (eg. FcγRII) in the former study, thereby dampening immune overactivation. More studies are needed to test this hypothesis and to identify how engagement of FcγRs can lead to either beneficial or destructive responses in experimental stroke.

2.2.2. FcγR signaling

As mentioned above, the Fc regions of immunoglobulins are well-known ligands for FcRs. In addition, several non-immunoglobulin ligands for FcRs have also been identified. For example, Galectin-3, an animal lectin, is an endogenous ligand for inhibitory FcγRII (Cortegano et al., 2000). The interaction between Galectin-3 and FcγRII inhibits the production of interleukin-5 (IL-5) from peripheral blood mononuclear cells. Interestingly, soluble forms of FcγRs (sFcγRIII) have been detected in human plasma (Fleit et al., 1992; Huizinga et al., 1990). These sFcRs can bind to non-immunoglobulin ligands, such as complement receptors on various cell types and induce cytokine production (Galon et al., 1996; Heyman, 2000). The biological significance of these non-immunoglobulins ligands has not been addressed for microglia or neurological disorders and warrants further investigation.

Most FcγRs are transmembrane glycoproteins. Their common α chain contains an extracellular region with varying numbers of the IgSF domain. The IgSF domain includes a transmembrane region and a cytoplasmic region. Some FcγRs have only a single α chain (eg. FcγRIIa and FcγRIIb). Others are coupled to additional receptor subunits (eg. two γ chains for FcγRI and FcγRIIIa) involved in downstream signaling (Ortiz-Stern and Rosales, 2003) (Figure 3).

Figure 3. Fcγ receptors.

Most FcγRs have a common α chain that contains an extracellular region with varying numbers of the immunoglobulin (Ig)-like domain. FcγRIIa has only a single α chain. FcγRI and FcγRIIIa are coupled to additional receptor subunits (two γ chains) that are involved in downstream signaling. The signaling pathways initiated by the engagement of different activating FcγRs usually begin with tyrosine phosporylation of the conserved immunoreceptor tyrosine-based activation motif (ITAM) in the cytoplasmic tail of the receptors or their associated subunits by kinases in the Src family. The subsequent recruitment of Syk-family kinases in turn permits interaction with other signaling proteins, such as PI3K, Src, and Ras, and further activates downstream signaling pathways that lead to specific biological functions. The activation of NFκB or the Ras/MEK/ERK pathway is critical for cytokine/chemokine production whereas the PI3K-Akt pathway is important for microglial phagocytosis. In contrast to the ITAM-mediated activation of microglia, the engagement of inhibitory FcγRIIb triggers immunoreceptor tyrosine-based inhibition motif (ITIM) activation through its cytoplasmic domain, resulting in the recruitment of phosphatases such as SHIP and SHP1. This attenuates the signaling cascades engaged by activating FcRs and diminishes downstream biological effects by acting in counterpoint to ITAM.

The signaling pathways initiated by the engagement of different activating FcγRs usually begin with tyrosine phosporylation of the conserved ITAM in the cytoplasmic tail of the receptors or their associated subunits by kinases in the Src family. The subsequent recruitment of Syk-family and/or zeta-chain-associated protein kinase 70 (ZAP70) allows for interactions with other signaling proteins, such as linker for activation of T cells (LAT), phosphoinositide-3-kinase (PI3K), Src homology 2 (SH2) domain containing leukocyte protein of 76 kDa, and PLCγ2; these interactions activate further downstream signaling pathways that lead to specific and divergent biological functions (Crowley et al., 1997; Gross et al., 1999; Jakus et al., 2009; Park and Schreiber, 1995; Saitoh et al., 2000; Taylor et al., 1997; Tridandapani et al., 2000). For example, Src and Syk kinases have been shown to be important for microglial phagocytosis and release of the chemokine MIP-1α following FcγR engagement (Song et al., 2004). Furthermore, activation of NFκB or the Ras/MEK/ERK pathway is critical for MIP-1α production whereas the PI3K-Akt pathway is important for microglial phagocytosis, suggesting that multiple downstream signaling pathways are activated by FcγR to regulate different functions (Figure 3). Thus, a single FcγR can initiate distinct microglial functions (eg. phagocytosis and cytokine production), thereby illustrating that pathway divergence is another important principle of microglial receptor engagement.

In contrast to the ITAM-mediated activation of microglia, the engagement of inhibitory FcγRIIb triggers ITIM activation through its cytoplasmic domain, resulting in the recruitment of phosphatases such as SHIP and SHP1. This attenuates the signaling cascades engaged by activating FcγRs and diminishes downstream biological effects by acting in counterpoint to ITAM (Huang et al., 2003) (Figure 3).

The coexpression of activating and inhibitory FcγRs on microglia and their recruitment of opposing signaling cascades upon binding the same ligand reveal how FcγRs may contribute to the maintenance of microglial equilibrium and balance various immune functions. It may also underlie the contradictory effects of FcγR engagement in stroke models (see above).

2.3. CD200 receptor (CD200R)

Cluster of Differentiation 200 (CD200) receptor (CD200R, also named the OX2 receptor) and its ligand CD200 (OX2) are both cell surface glycoproteins that contain two immunoglobulin domains. While CD200 is broadly expressed on many types of cells including lymphoid cells, neurons and endothelial cells, CD200R is exclusively expressed on leukocytes, especially cells of the myeloid lineage (Wright et al., 2000).

2.3.1. CD200R maintains microglia in a quiescent state

CD200R was first identified on macrophages (Preston et al., 1997) and later detected on microglia (Wright et al., 2000). CD200R actively interacts with CD200 on healthy neurons and maintains microglia in a quiescent state. Disruption of CD200-CD200R binding with CD200 blocking antibodies increases microglial activation and exacerbates the signs of experimental autoimmune encephalomyelitis (EAE) (Wright et al., 2000) and experimental Parkinson’s disease (PD) (Zhang et al., 2011b) in rats. Studies in CD200 deficient mice further confirm that, in the absence of the inhibitory signaling from neuronal CD200, microglia spontaneously exhibit features of activation, including enlarged cell bodies, shortened and thickened processes, and increased expression of CD11b and CD45 (Hoek et al., 2000). Similar to the results with CD200 antibodies, CD200 deficiency also accelerates the onset of and enhanced microglia/macrophage activation in EAE mice (Copland et al., 2007; Hoek et al., 2000). These studies strongly suggest that CD200R is an important inhibitory receptor on microglia and that it plays a major role in maintaining immune balance in the CNS.

The involvement of CD200-CD200R in acute CNS injuries has also been studied. For example, in facial nerve axotomy, a traditional model for in situ microglial activation induced by damaged neurons, microglial activation is more apparent in CD200-null mice than in control mice (Hoek et al., 2000). This study shows that CD200 suppresses microglial activation also in response to acute neuronal injury. Furthermore, in the MCAO model of ischemic stroke, CD200 mRNA levels are significantly decreased as early as 1 day after ischemia (Matsumoto et al., 2007). Levels of CD200 mRNA plateau at 5 days and gradually increase back to control values thereafter. This acute post-ischemic drop in CD200 levels may contribute to short-term microglial activation after stroke. Interestingly, macrophage-like cells in the ischemic core have been shown to express CD200, suggesting that CD200-CD200R engagement may also be involved in the crosstalk between microglia and other immune cells.

2.3.2. CD200R signaling

CD200 is the only known ligand for CD200R. CD200R interacts with CD200 through its N-terminal Ig-like domain (Hatherley and Barclay, 2004). Unlike most other inhibitory receptors, CD200R does not contain any ITIM, nor does it recruit any tyrosine phosphatases upon engagement (Mihrshahi and Brown, 2010). Instead, CD200R has a cytoplasmic phosphotyrosine-binding (PTB) domain recognition motif (NPxY). Phosphorylation of the tyrosine residue within this motif leads to the direct binding of this motif to the PTB domain-containing adaptor downstream of tyrosine kinase 2 (Dok2) (Mihrshahi and Brown, 2010; Zhang et al., 2004; Zhang and Phillips, 2006). Further phosphorylation of CD200R-bound Dok2 recruits and activates Ras GTPase activating protein (RasGAP). RasGAP hydrolyzes RasGTP into the inactive form RasGDP, resulting in the subsequent inhibition of PI3K and ERK signaling (Mihrshahi et al., 2009). In addition to this main inhibitory pathway, CD200R ligation also causes a relatively delayed phosphorylation of Dok1 (Mihrshahi et al., 2009; Mihrshahi and Brown, 2010; Zhang et al., 2004). Upon phosphorylation, Dok1 activates other signaling molecules and finally inhibits the activation of Dok2 and RasGAP (Mihrshahi and Brown, 2010). Therefore, the activation of Dok1 initiates an elegant negative feedback cycle to fine-tune the activation of CD200R signaling pathway.

2.4. Receptor for advanced glycation endproducts (RAGE)

RAGE is named after its ability to bind advanced glycation end-products, which are proteins or lipids that are non-enzymatically glycated after exposure to sugars. RAGE exists in membrane-bound forms or, alternatively, in soluble forms that lack the transmembrane domain (Hudson et al., 2003). RAGE is highly expressed on immune cells and contributes to many pathological states with inflammatory components (Sparvero et al., 2009).

2.4.1. RAGE mediates microglia accumulation and activation in CNS injuries

RAGE is highly expressed in the CNS. During early development, RAGE is mainly found in neurons and contributes to neurite outgrowth (Hori et al., 1995). In the adult CNS, RAGE is expressed in various types of CNS cells including microglia, neurons, and endothelial cells (Giri et al., 2000; Yan et al., 1996). The function of microglial RAGE was initially characterized in models of AD. In postmortem tissue from AD victims, RAGE immunoreactivity is increased within microglia in close proximity to senile plaques (Lue et al., 2001; Yan et al., 1996). Further in vitro studies have confirmed that RAGE is present in cultured microglia and serves as a direct receptor for Aβ (Yan et al., 1996). The involvement of microglial RAGE in brain injuries has also been reported. In in vitro and in vivo models of ischemic stroke, high-mobility group box 1 (HMGB1), a ligand for RAGE, is released from injured neurons early after ischemia and contributes to ischemic brain injury through its interaction with RAGE (Muhammad et al., 2008). Further studies using chimeric mice generated by transplanting RAGE−/− bone marrow into wild-type mice confirmed that RAGE expression on immigrant macrophages was essential for post-stroke cerebral inflammation and brain damage. This study also suggests that HMGB1-RAGE signaling links neuronal necrosis with microglia/macrophage activation. Thus, RAGE signaling is a rational target for anti-inflammatory therapy in stroke. In animal models of TBI and in TBI patients, upregulation of RAGE in the brain has been reported, with a prominent increase in phagocytic microglia surrounding the lesion core at late stages of injury. Indeed, HMGB1 is released from cells early after TBI and primarily expressed in phagocytic microglia at late stages of TBI (Gao et al., 2012). However, further studies are needed to elucidate the biological significance of the temporal changes in RAGE and HMGB1 in TBI.

2.4.2. RAGE signaling

RAGE is an integral membrane protein composed of an extracellular domain, a single transmembrane-spanning domain, and a short, highly charged cytosolic domain (Hudson et al., 2008). RAGE recognizes many different binding partners through a common motif and is therefore also known as a pattern recognition receptor. Microglial RAGE has been documented to recognize ligands such as advanced glycation endproducts (AGEs) (Dukic-Stefanovic et al., 2003), Aβ (Yan et al., 1996), HMGB1 (Muhammad et al., 2008), and S100 proteins (Bianchi et al., 2007; Bianchi et al., 2010).

Ligand binding to microglial RAGE triggers many different intracellular signaling pathways. Microglial RAGE signaling following Aβ binding has been the most thoroughly investigated. Aβ-RAGE interactions elevate microglial macrophage colony-stimulating factor (m-csf) expression and initiate microglial chemotaxis (Lue et al., 2001). In addition, RAGE enhances secretion of several proinflammatory mediators (eg. TNF-α, IL-1β) from microglia, partly through the activation of p38 and ERK1/2. Such RAGE-induced neuroinflammation has the potential to greatly exaggerate Aβ neurotoxicity, leading to impaired learning and memory (Fang et al., 2010). Furthermore, IL-1β, p38MAPK and JNK are all involved in microglial RAGE signaling and mediate Aβ-induced synaptic dysfunction and impairments in long-term potentiation (Origlia et al., 2010).

RAGE activation in microglia by AGEs results in the upregulation of pro-inflammatory cytokines such as IL-6, TNF-α and iNOS through the activation of NFκB, MEK, and PI3K signaling pathways (Dukic-Stefanovic et al., 2003). AGE-RAGE interactions have been shown to induce microglia inducible nitric oxide synthase (iNOS) expression in a JNK-dependent manner and to elicit subsequent NO production (Khazaei et al., 2008). AGEs have also been found to stimulate the expression of RAGE in a human microglial cell line. AGE-RAGE interactions induce TNF-α and TNF-α receptor II expression on microglia, which in turn enhance the expression of the gap junction protein connexin 43 and thereby enhance communication between activated microglia (Shaikh et al., 2012).

S100B stimulates NO production from microglia in a manner dependent on RAGE extracellular domains but independent of RAGE transducing activity (Adami et al., 2004). These findings suggest that RAGE may concentrate S100B on the surface of microglia. In contrast, S100B-induced upregulation of cyclo-oxygenase-2 (COX2) expression in microglia seems to be RAGE-dependent through simultaneous stimulation of the Cdc42-Rac1-JNK and Ras-Rac1-NF-κB pathways (Bianchi et al., 2007). In addition, microglial RAGE engagement by S100B induces IL-1β and TNF-α expression by concurrent stimulation of NFκB and AP-1 transcriptional activities (Bianchi et al., 2010). A recent study also documented a role for S100B-engaged RAGE in microglial migration via activation of multiple signaling pathways and the resulting upregulation of several chemokines (CCL3, CCL5, and CXCL12) and chemokine receptors (CCR1 and CCR5) (Bianchi et al., 2011). These results suggest that S100B engages the RAGE receptor to turn on microglial function. However, the S100B-RAGE interaction does not always turn on microglial function. For example, in special pathological conditions such as tumor formation, S100B actually attenuates the activation of glioma associated microglia and macrophages, through the induction of STAT3 (Zhang et al., 2011a). As a result of these findings, blunt suppression of all the effects of RAGE in the clinic might lead to unwanted side effects and investigators will have to work towards a subtler fine-tuning of its pro- and anti-inflammatory properties.

3. Chemokine receptors

Chemokines belong to a family of small secreted proteins with molecular mass usually less than 10 kDa. Chemokines are so-named because they are leukocyte chemoattractants and also possess cytokine activities, in a portmanteau of the words ‘chemoattractant’ and ‘cytokine’ (Asensio and Campbell, 1999). Chemokines have four conserved cysteine residues that form disulfide bonds. The chemokine family is divided into four subclasses (CC, CXC, CX3C and C subclass) according to the position of the first two cytokines in the N-terminal region. Chemokines send signals through chemokine receptors, all of which are G-protein-coupled. The interaction between chemokines and chemokine receptors not only directs immune cell migration, but also controls a variety of other cellular functions. The receptors most often implicated in microglial functions in CNS injuries are CX3CR1, CCR2, CXCR4, CCR5, and CXCR3.

3.1. CX3CR1

3.1.1. CX3CR1 and its ligand CX3CL1 regulate microglia-neuron crosstalk

CX3CR1 is mainly expressed on glial cells in the CNS, especially microglia (Harrison et al., 1998; Meucci et al., 1998; Nishiyori et al., 1998). CX3CR1 exhibits a monogamous relationship (Imai et al., 1997) with its ligand CX3CL1 and plays an important role in neuron-microglial crosstalk. CX3CL1 (also named fractalkine) was the first chemokine found to be expressed in neurons (Harrison et al., 1998; Meucci et al., 1998; Nishiyori et al., 1998). Its expression in the brain is higher than in the periphery and mainly localized to the olfactory bulb, cerebral cortex, hippocampus, caudate, putamen, and nucleus accumbens (Nishiyori et al., 1998). CX3CL1 can either be membrane-bound or released from the cell surface after constitutive or inducible cleavage by proteases of the A Disintegrin And Metalloprotease (ADAM) family (Garton et al., 2001; Hundhausen et al., 2003). CX3CL1 is one of the ‘find me’ signals that apoptotic cells utilize to advertise their presence and to recruit local phagocytes (Ravichandran, 2011). Thus, CX3CL1 is important for the adaptive clearance and removal of dysfunctional, dying cells from the organism.

Mechanistic studies on CX3CR1 have revealed that it modulates multiple microglial activities, including adhesion, migration, proliferation, inflammatory cytokine production, and clearance capacity (Bhaskar et al., 2010; Cardona et al., 2006; Denes et al., 2008; Lyons et al., 2009). Importantly, CX3CR1 inhibits the microglial production of inflammatory cytokines following stimulation (Bhaskar et al., 2010; Cardona et al., 2006; Denes et al., 2008). The role of CX3CR1 in microglial phagocytosis is still controversial. Some studies have documented that CX3CR1 signaling inhibits or has no effect on microglial phagocytosis of β-amyloid in models of AD (Lee et al., 2010; Liu et al., 2010). However, other studies show that stimulation of CX3CR1 with its ligand increases microglial clearance of neuronal debris and induces neuroprotection (Noda et al., 2011). These differences might reflect the varying roles of CX3CR1 in the presence of distinct stimuli. Further work to distinguish how CX3CR1 exerts these potentially dualistic roles is warranted.

3.1.2. CX3CR1 and CX3CL1 in CNS injuries

The biological functions of CX3CL1 and CX3CR1 signaling in vivo have been investigated using CX3CR1-deficient mice. Under physiological conditions, disruptions in CX3CR1 signaling lead to impairments in cognitive function and synaptic plasticity via increased action of IL-1β (Rogers et al., 2011). These studies indicate an important role for CX3CR1 in maintaining CNS homeostasis and normal interneuronal communication. CX3CR1 deficiency augments microglial activation and neuronal damage in models of neurodegenerative diseases such as PD, AD, amyotrophic lateral sclerosis, and macular degeneration (Cardona et al., 2006; Chen et al., 2007; Cho et al., 2011; Pabon et al., 2011), suggesting an inhibitory effect of CX3CR1 on inappropriate microglial activation and subsequent neuronal toxicity. Paradoxically, however, CX3CR1 knockout mice display reduced neuroinflammation and mitigated neuronal damage after focal cerebral ischemia and SCI (Denes et al., 2008; Donnelly et al., 2011). A recent study provided a plausible explanation for this discrepancy (Donnelly et al., 2011):-in experimental paradigms where microglia are the predominant CNS macrophages (e.g., PD and amytrophic lateral sclerosis), deficient CX3CR1 signaling was proposed to enhance microglial neurotoxicity due to the loss of inhibitory signaling through CX3CR1-CX3CL1 interactions (Cardona et al., 2006). In contrast, in acute neurological injuries such as stroke or spinal cord injury, early microglial activation is followed by prominent macrophage infiltration, which contributes greatly to the expansion of primary injuries. A deficiency in CX3CR1 may limit the recruitment and activation of peripheral macrophages and thus exert neuroprotective effects. Therefore, CX3CL1-CX3CR1 signaling may exert opposite effects in different pathological conditions.

Another factor that may complicate the impact of CX3CL1-CX3CR1 is the timing of the immune response in relation to disease progression. For example, increased CX3CL1-CX3CR1 signaling may be beneficial in early AD by restraining microglial activity and preventing the exacerbation of phosphorylated tau-mediated damage to neurons (Bhaskar et al., 2010; Cho et al., 2011). However, sustained elevation of CX3CL1-CX3CR1 signaling may become detrimental by interfering with microglial phagocytosis in advanced stages when removal of extracellular Aβ deposits becomes necessary (Desforges et al., 2012; Lee et al., 2010). All of these complex issues need to be taken into full consideration when developing therapeutic interventions targeting CX3CR1.

Beyond the abovementioned regulation of inflammatory responses, CX3CL1 may sensitize microglia in the spinal cord and cause allodynia and thermal hyperalgesia. These additional functions of CX3CL1 have been proposed because antibodies against CX3CR1 are known to inhibit the development of neuropathic pain (Clark et al., 2007; Milligan et al., 2004; Zhuang et al., 2007). Further studies have also shown that membrane-bound CX3CL1 can be cleaved by microglial cathepsin S after SCI. The released soluble CX3CL1 then activates microglia through an interaction with CX3CR1 and causes neuropathic pain (Clark et al., 2007; Clark et al., 2009). These data provide potential therapeutic targets on microglia for conditions involving neuropathic pain.

3.1.3. CX3CL1-CX3CR1 signaling

CX3CR1 is a G-protein-coupled seven transmembrane domain receptor. Upon CX3CL1 recognition and binding, CX3CR1 undergoes a conformational change and activates associated G proteins by exchanging bound GDP for GTP, known as guanine nucleotide exchange. Following this exchange, the α subunit of the G-protein dissociates from the β/γ subunits to activate other intracellular signaling proteins, including phospholipase C (PLC), MAPK, and PI3K (Savarin-Vuaillat and Ransohoff, 2007). For example, the engagement of CX3CR1 on monocytes has been shown to induce the coordinated activation of ERK, p38, JNK1, and PI3K to efficiently mediate the CX3CR1-dependent cell adhesion to the extracellular glycoprotein fibronectin (Cambien et al., 2001). Thus, chemokine receptors play a role in cell adhesion to the extracellular matrix, which is known to regulate cell migration and is consistent with the chemoattractive properties of their cytokine ligands.

Despite the chemotactic properties of CX3CL1-CX3CR1 interactions in monocytes, CX3CL1-CX3CR1 interactions actually maintain microglia in a quiescent state by the activation of PI3K, as manifested by reduced IL-1 production after LPS exposure (Lyons et al., 2009). It has also been shown that the activation of the p38 pathway by CX3CR1 is important for pain processing after spinal cord injury or bone cancer (Hu et al., 2012; Zhuang et al., 2007). Apparently, multiple signaling pathways are activated downstream of CX3CL1-CX3CR1 engagement and are involved in a variety of cellular functions. The specific signaling pathways activated by CX3CR1 still need to be explored for each individual function, such as phagocytosis and cytokine production.

3.2. CCR2

3.2.1. CCR2 and its ligands direct microglial migration toward the site of damage

The chemokine receptor CCR2 is widely expressed on almost all types of immune cells including lymphocytes, neutrophils, macrophages, and natural killer cells (Savarin-Vuaillat and Ransohoff, 2007). As expected from its classification as a chemokine receptor, the main function of CCR2 is to direct immune cells to the site of inflammation, where the CCR2 ligands accumulate (Serbina and Pamer, 2006; Tsou et al., 2007). The chemoattractant functions of CCR2 are important for monocyte deployment from bone marrow or other lymph organs and for their subsequent infiltration into the CNS under pathological circumstances (Babcock et al., 2003; Bao et al., 2010; Chen et al., 2001; D’Mello et al., 2009; Eugenin et al., 2006; Naert and Rivest, 2012). It has been shown that circulating Ly-6C^high ‘inflammatory’ monocytes equipped with CCR2 can penetrate the brain upon blood-brain barrier disruption and serve as migratory precursors of microglia (Getts et al., 2008; Mildner et al., 2011; Mildner et al., 2007). The constitutive expression of CCR2 on resident microglia is very low if present. However, LPS and other pathological stimuli dramatically induce the expression of this receptor on microglia (Banisadr et al., 2002b; Boddeke et al., 1999; Sivakumar et al., 2011; Zhang et al., 2007). In this manner, microglia might acquire migratory properties only in the presence of the appropriate inflammatory stimuli.

3.2.2. Microglial CCR2 and its ligands in CNS injuries

Unlike CX3CR1, CCR2 has multiple ligands, namely CCL2 (also named monocyte chemoattractant protein (MCP)1), CCL7 (MCP3), CCL8 (MCP2), CCL12 (MCP5), CCL13 (MCP4) and CCL16 (Charo et al., 1994; Franci et al., 1995; Garcia-Zepeda et al., 1996; Moore et al., 1997; Nomiyama et al., 2001). CCR2 and/or its ligands are upregulated in many types of CNS injury, including ischemia, hemorrhage, trauma, and hypoxia (Deng et al., 2009; Dimitrijevic et al., 2007; Kim et al., 2008). CCR2 activation on microglia after CNS injuries was initially reported to be destructive. This was based on findings that depletion of CCL2 or CCR2 reduces microglial activation and neuronal degeneration following aspiration lesions of the visual cortex in adult mice (Muessel et al., 2002). Similarly, it was shown that ischemia-induced infarcts are reduced in CCR2-deficient mice and that this is accompanied by reduced cerebral inflammation (Dimitrijevic et al., 2007). These studies suggest that the CCL2-CCR2 axis plays manifold roles in cerebral ischemia, including recruiting peripheral leukocytes and disrupting the blood-brain barrier. Subsequent studies using green fluorescent protein (GFP)-transgenic bone marrow chimeras further demonstrated that CCR2 deficiency reduces the influx of GFP-positive cells (macrophages and neutrophils) into the brain after ischemic stroke, whereas the activation and migration of resident microglia was not affected (Schilling et al., 2009)). It seems plausible that CCR2 is more important in peripheral immune cell trafficking than in local microglial responses following CNS injuries and that activation of CCR2 leads to the secondary expansion of primary damage.

In contrast, some studies suggest a possible protective role of the CCL2-CCR2 system in brain recovery. For example, although a lack of CCL2 or CCR2 decreases hematoma volume early after collagenase-induced intracerebral hemorrhage (ICH), it also delays the late phase resolution of the hematoma (Yao and Tsirka, 2012). This study also found that microglia activation/migration was attenuated early after ICH in CCR2 or CCL2 knockout mice and that this was followed by an exaggerated activation of microglia in later phases, possibly due to the activation of alternative signaling pathways. Therefore, any therapeutic strategies designed to target the CCR2-CCL2 complex after CNS injuries must be carefully timed to avoid adverse effects on long-term recovery and compensatory activation of alternative pro-inflammatory signalling.

Similar to CX3CR1, microglial CCR2 has been shown to be important in inflammatory pain and neuropathic pain states (Abbadie et al., 2003). To distinguish the roles of central and peripheral CCR2, bone marrow chimeric mice have been constructed to achieve selective knockout of CCR2 in resident microglia or in bone-marrow derived macrophages (BMDM). The results show that CCR2 expression in either resident microglia or BMDM is sufficient for the development of mechanical allodynia after peripheral nerve injury induced by partial ligation of the sciatic nerve (Zhang et al., 2007). These data support an essential role for microglial receptors in the development of neuropathic pain and reveal a potential therapeutic target for conditions associated with inflammation.

3.2.3. CCR2 signaling

A model of neuron-microglia interactions through the CCL2-CCR2 system has recently been proposed by Yao and colleagues (Yao and Tsirka, 2010). In their model, the C-terminus of CCL2, which is extensively decorated with O-linked carbohydrates, anchors the CCL2 molecule on neurons through its interaction with the glycosaminoglycans (GAGs) on neuronal cell membranes. This event increases the local CCL2 concentration, which in turn promotes the formation of CCL2 dimers or higher order aggregates. The C-terminal extension negatively regulates the chemotactic ability of CCL2 and hinders CCL2-CCR2 crosstalk (Sheehan et al., 2007). Upon neuronal injury or other noxious stimuli, plasmin is released from disturbed neurons and cleaves the C-terminal extension of CCL2 and abrogates CCL2 dimers. The truncated CCL2 N-terminal monomers then bind to CCR2 on microglia and initiate microglial intracellular signaling pathways. However, it was also noted that this model is based on mouse CCL2 structure. Human CCL2 does not have a glycosylated C-terminal, although its N-terminal is highly homologous to mouse CCL2. Therefore, other mechanisms might have evolved to regulate CCR2-CCL2 interactions in higher species (Yao and Tsirka, 2010).

The engagement of CCR2 induces GDP/GTP exchange on the α subunit (αi, αq or α16) of the associated G protein (Arai and Charo, 1996). The α–GTP complex then dissociates from the βγ heterodimer, and both complexes activate intracellular signaling cascades that differ depending on cell type. For example, the αi–GTP complex inhibits adenylate cyclase and the βγ complex activates phospholipase C to generate diacylglycerol and inositol 1,4,5-trisphosphate, leading to the release of intracellular calcium and activation of protein kinase C (PKC) (Myers et al., 1995). Activated CCR2 can also bind directly to FROUNT, a unique clathrin heavy-chain repeat homology protein, and form clusters at the leading edge of migrating cells (Terashima et al., 2005). Association of FROUNT with activated CCR2 leads to the activation of PI3K-Rac (a member of the Rho family of small GTPase signaling proteins) and the resultant protrusion of lamellipodia, the cytoskeletal actin projections observed at the leading edges of migrating cells. Although the involvement of FROUNT in microglial migration is not yet definitive, the activation of Rac is known to be important for the protrusion of lamellipodia and microglial migration upon CCR2 stimulation (Yao et al., 2010).

3.3. CCR5

3.3.1. Microglial CCR5 and its ligands in CNS injuries

CCR5 is expressed on many types of leukocytes involved in immune regulation. Like all the other chemokine receptors, CCR5 is important for immune cell migration toward specific ligands and thus results in recruitment of leukocytes to the site of inflammation or injury (Samson et al., 1996). The natural chemokine ligands that bind to CCR5 include CCL5 (also named regulated and normal T cell expressed and secreted, RANTES), CCL3, and CCL4. The importance of CCR5 in microglia was first reported more than a decade ago when it was identified as one of the most important co-receptors for HIV infection of microglia (He et al., 1997; Shieh et al., 1998). Subsequent studies revealed that CCR5 is expressed in many types of CNS cells (microglia, neurons, astrocytes, and vascular endothelial cells) and plays multifarious roles mediating glial-glial and glial-neuronal interactions (Cartier et al., 2003; Gamo et al., 2008; Rottman et al., 1997; Skuljec et al., 2011).

CCR5 is expressed at low levels by microglial cells throughout normal CNS tissue. In contrast, high expression of microglial CCR5 has been observed in several neurological injuries including excitotoxic brain injury (Galasso et al., 1998), perinatal hypoxia-ischemia (Cowell et al., 2006) and TBI (Carbonell et al., 2005). The primary function of upregulated CCR5 in neurodegenerative diseases is to direct microglial migration (Marella and Chabry, 2004). In contrast, the functions of microglial CCR5 are more complicated in acute CNS injuries. For example, in an experimental model of focal brain injury, a specific CCR5 antagonist, TAK-779, was able to reduce migration velocity and the number of perilesional migratory microglia (Carbonell et al., 2005) highlighting an additional trafficking role for CCR5. However, the upregulation of CCR5 following acute motor nerve injury is not essential for the recruitment of microglia. This is evidenced by the normal migration of microglia toward injured motor neurons in CCR5 deficient mice (Babcock et al., 2003; Gamo et al., 2008). An alternative function of CCR5 may be to protect neurons by ameliorating the production of NO and inflammatory cytokines from microglia following initial nerve damage (Gamo et al., 2008). However, conflicting in vitro findings also show that the activation of CCR5 induces a pro-inflammatory profile in microglia, as manifested by increased NO and reduced IL-10 production following CCL5 exposure (Skuljec et al., 2011). Further studies to explain the distinct functions of CCR5 on microglial behavior in different experimental paradigms are warranted.

3.3.2. CCR5 signaling

Signaling pathways initiated by G protein coupled chemical receptors are frequently related to ion channel regulation (Dascal, 2001). It is therefore not surprising that some microglial receptors open ion channels. For example, CCR5 is coupled to pertussis toxin (PTX)-sensitive inhibitory G protein (GαIβγ) and exerts its cellular function by regulating intracellular Ca²⁺ concentrations (Shideman et al., 2006). The engagement of CCR5 results in the dissociation of βγ subunits from the Gi heterotrimer. The liberated βγ subunits activate multiple signaling pathways, including PI3K/PLC/Bruton’s tyrosine kinase (Btk)/Cyclic ADP-ribose (cADPR)-mediated Ca²⁺ release from intracellular stores and ADP-ribose (ADPR)-dependent Ca²⁺ influx through nimodipine-sensitive Ca²⁺ channels on the plasma membrane.

3.4. CXCR3

mRNA expression of CXCR3 was first reported in cultured microglia by Biber and colleagues using RT-PCR (Biber et al., 2002) and was confirmed with in situ hybridization, immunohistochemistry, and flow cytometry (Biber et al., 2001; Flynn et al., 2003). CXCR3 binds to multiple CXC chemokines including CXCL4 (also named Platelet factor 4 (PF4)), CXCL9 (also named monokine induced by gamma interferon (MIG)), CXCL10 (also named interferon gamma-induced protein 10 (IP10)), CXCL11 (also named IP9); and CXCL21 (also named secondary lymphoid-tissue chemokine (SLC)) (Clark-Lewis et al., 2003; de Jong et al., 2008; Rappert et al., 2002; van Weering et al., 2011). Interestingly, many of these chemokines, such as CXCL21 and CXCL10, are markedly induced in injured neurons (Biber et al., 2001). The interaction between microglial CXCR3 and its ligands on endangered neurons is important for neuron-microglia crosstalk and the regulation of neuroinflammation in response to neuronal damage. Furthermore, these neuronal chemokines can be packed into vesicles and transported throughout neuronal processes to reach distant presynaptic structures (de Jong et al., 2005). This feature may explain the activation of microglia at locations remote from primary lesion sites. The major biological effect of microglial CXCR3 is to direct microglial migration towards CXCR3 ligands (Biber et al., 2001; Dijkstra et al., 2004; Rappert et al., 2004). In addition, microglial CXCR3 activation may attenuate microglial phagocytosis and production of NO upon LPS stimulation (de Jong et al., 2008). It is not known whether the activation of microglial CXCR3 leads to neuronal rescue or further neuronal damage. Nevertheless, a recent study suggests that the CXCR3+ microglia is neurosupportive, displaying high sensitivity to apoptosis and permitting the production of high levels of IL-10 and TGF-β (Li et al., 2006). Of note, CXCR3 function also depends on the cellular background on which the receptor is expressed because human CCL21 does not show any effects in CXCR3-transfected HEK293 cells, which are of human origin (Dijkstra et al., 2004).

CXCR3-mediated microglia-neuron interactions are known to be involved in brain injuries. Neuronal expression of CXCL21 is induced in the brain soon after the onset of ischemia, suggesting an involvement of CXCL21 and CXCR3 in early neuroimmunological events after stroke (Biber et al., 2001). Rappert and colleagues have examined the impact of CXCR3 on microglial migration and proliferation using the entorhinal cortex lesion and facial nerve axotomy models, respectively (Rappert et al., 2004). The authors found that CXCR3 signaling was critical for the migration but not for the proliferation of microglia after brain injuries (Biber et al., 2001; Rappert et al., 2002). These findings were consistent with their in vitro studies using microglial cultures. Interestingly, the impaired microglial migration after entorhinal cortex lesions in CXCR3 knockout mice was accompanied by the preservation of denervated dendrites. Denervated dendrites that have lost their afferent synaptic contacts are ordinarily cleared by microglia. Thus, the studies by Rappert and colleagues suggest that CXCR3 is essential for microglial clearance of dysfunctional dendrites. A recent study further suggests that microglial CXCR3 may regulate neuronal death in a brain region-specific or topographic manner (van Weering et al., 2011).

Very little information is available on the signal transduction pathways activated by microglial CXCR3 and their possible correlation with the biologic actions elicited by agonists. The engagement of CXCR3 on different types of cells has been associated with the activation of a variety of signaling pathways, including the p38, ERK, Src, PI3K/Akt, and PLC pathways (Bonacchi et al., 2001; Shahabuddin et al., 2006; Smit et al., 2003). The involvement of these pathways in microglial CXCR3 signaling, however, has not been investigated. Nevertheless, activation of microglial CXCR3 is known to be associated with the opening of ion channels. Thus, brief exposure of microglia to CCL21 is known to trigger a long lasting elevation in Cl⁻ conductance, which is critical for CCL21-induced chemotaxis (Rappert et al., 2002). Further identification of CXCR3 signaling will shed more light on the function of this chemokine receptor in microglia.

3.5. CXCR4 and CXCL12

CXCR4 is constitutively expressed in the CNS in a variety of cell types including microglia, astrocytes, neurons, oligodendrocyte progenitors, and vascular endothelial cells (Banisadr et al., 2002a; Ohtani et al., 1998; Patel et al., 2010). The CXCR4 receptor is coupled with GαIβγ protein. PTX, which specifically uncouples the Gαi subunit from G protein, can inhibit CXCL12-induced calcium mobilization in microglia (Tanabe et al., 1997). CXCL12 (also named stromal-derived-factor-1, SDF-1) was initially identified as the only ligand for CXCR4. However, recent studies demonstrate that ubiquitin is also a natural ligand of CXCR4 (Saini et al., 2010). Although many studies on CXCR4 function in CNS focus on its involvement in HIV infection, accumulating evidence suggests that the CXCR4-CXCL12 axis has important functions in normal physiology in addition to an important role in disease states (Limatola et al., 2000; Zou et al., 1998). In support of this notion, mice with either CXCR4 or CXCL12 gene deletions die soon after birth with severe abnormalities affecting many organs, including the brain. Studies by Bezzi and colleagues further demonstrate that the CXCR4-CXCL12 system is important for communication between astrocytes and neurons via astrocytic glutamate release (Bezzi et al., 2001). However, excessive CXCL12 in the injured or infected brain activates CXCR4 on microglia, which then release large amounts of TNFα. This massive release of TNFα results in turn in a surge of glutamate release from astrocytes. These responses switch astrocyte function from normal physiology to excitotoxicity, perturbing the transmission of information from glial networks to neuronal circuits. CXCR4 engagement has also been shown to regulate IL-6 production from microglia in an ERK, PI3K, and NFκB-dependent manner (Lu et al., 2009). As expected from its classification as a chemokine receptor, CXCR4 also mediates microglial migration toward CXCL12 (Wang et al., 2008).

CXCL12 and CXCR4 are implicated in many CNS injuries. For example, CXCL12 and CXCR4 are involved in the pathology of ischemic stroke. Cerebral ischemia results in a prompt and long-lasting elevation of CXCL12 in the ischemic penumbra, particularly in association with reactive astrocytes in perivascular areas (Hill et al., 2004). Interestingly, transplanted GFP+ bone marrow cells are recruited in proximity to these CXCL12+ vessels and display the characteristics of activated microglial cells. These results suggest that CXCL12 is important in the homing of bone marrow-derived monocytes, which transform into microglia at the site of ischemic injury.

An in vitro study with microglial cultures suggests that exposure to hypoxia significantly enhances CXCR4 expression on microglia in a hypoxia inducible factor-1alpha (HIF-1alpha)-dependent manner, resulting in the accelerated migration toward CXCL12 (Wang et al., 2008). Such accumulation of monocyte-derived and local microglia may contribute to severe neuroinflammation after brain injury. However, there is also evidence that CXCL12 and CXCR4 might be important for neurorepair processes after brain ischemia by regulating neurogenesis and angiogenesis within the vascular niche (Merino et al., 2011; Wang et al., 2012). Finally, chronic intrathecal delivery of CXCL12 after spinal cord contusion has been shown to reduce cell apoptosis, boost astroglial and microglial responses, enhance angiogenesis, and ultimately improve long-term functional recovery after SCI (Zendedel et al., 2012). Such dualistic roles of CXCL12/CXCR4 in neuroinflammation and neurovascular repair need to be carefully considered when targeting this molecular pair with rational drug design.

4. Purinergic receptors

Purinergic receptors (purinoreceptors) are divided into two broad classes, P1 and P2, based on their binding properties. P1 receptors bind adenosine whereas P2 receptors bind ATP.

4.1. P1 adenosine receptors

P1 adenosine receptors are a class of G protein coupled seven-transmembrane proteins. Four different subtypes of adenosine receptors have been identified: A1, A2A, A2B, and A3. Each receptor subtype is encoded by a separate gene and exerts both distinct and overlapping functions. In general, A1 and A3 receptors are coupled to Gi and mediate biological functions distinct from those mediated by Gs protein-coupled A2A or A2B receptors. Purine nucleoside adenosine is the natural ligand of P1 adenosine receptors. The concentration of extracellular adenosine is very low under physiological conditions. This small amount of adenosine nonetheless exerts important roles in the modulation of many brain functions, including sleep and arousal, locomotion, anxiety, cognition and memory (Hasko et al., 2005). Under pathological CNS conditions such as hypoxia, ischemia, or trauma, adenosine may be released from injured neurons and other types of cells as a consequence of energy metabolism failure (Headrick et al., 1994; Koos et al., 1997; Latini et al., 1996). This release results in both protective and injurious effects on tissue via the stimulation of various P1 adenosine receptors.

4.1.1. A2A receptor

All four subtypes of adenosine receptors have been detected on microglia (Hasko and Pacher, 2012; Hasko et al., 2005; van der Putten et al., 2009) and were shown to individually or cooperatively modulate microglial activities such as proliferation, apoptosis, protein production, and chemotaxis. Among these receptors, A2A appears to be actively involved in the secretion of inflammatory factors from stimulated microglia. The expression of microglial A2A receptors is enhanced with inflammation, such as following stimulation with LPS or other toll-like receptor (TLR) ligands (van der Putten et al., 2009; Wittendorp et al., 2004). Microglial A2A receptors are also increased in pathological conditions such as AD and brain ischemia (Angulo et al., 2003; Trincavelli et al., 2008). In vitro experiments reveal that A2A receptor activation potentiates the release of NO from LPS-stimulated microglia (Saura et al., 2005) and leads to the upregulation of COX2 and the release of prostaglandin E2 (PGE2) from cultured microglia (Fiebich et al., 1996). These results may explain the anti-inflammatory and neuroprotective effects of A2A antagonists or gene knockout in animal models of cerebral ischemia, TBI, and SCI (Chen et al., 1999; Genovese et al., 2009; Yu et al., 2004).

A recent elegant study proposed that local glutamate levels may dictate the function of A2A receptors in microglia. In the presence of low concentrations of glutamate, an A2A receptor agonist inhibited LPS-induced microglial activation. High concentrations of glutamate, however, redirected A2A receptor signaling from the PKA to the PKC pathway, resulting in a switch from anti-inflammatory to pro-inflammatory effects (Dai et al., 2010). Therefore, the neuroprotective effect of A2A antagonists and agonists may well depend on the extent of brain injury and disease stage. Therapies targeted at adenosine pathways must take this into full consideration.

In addition to protein factor secretion, the A2A receptor is also involved in other microglial functions. For example, the A2A receptor is important for the retraction of microglial processes, which results in the amoeboid morphology assumed by activated microglia during in vivo LPS exposure or brain injury. Finally, the A2A receptor has also been shown to promote microglial proliferation and inhibit phagocytosis by LPS-activated microglia (Gebicke-Haerter et al., 1996). The contribution of these A2A functions to CNS pathologies awaits further investigation.

4.1.2. A3 receptor

The function of A3 receptors on microglia is not clear. Conflicting results have been presented by different groups. The early studies of microglial A3 receptors documented that stimulation of A3 receptors activates Gi protein-dependent phosphorylation of ERK1/2 and p38. However, the role of the A3 receptor in microglial function was not addressed (Hammarberg et al., 2004; Hammarberg et al., 2003). A subsequent report argued that the A3 receptor had no impact on the phosphorylation of p38 and ERK1/2. Instead, the A3 receptor was reported to inhibit LPS-induced PI3K/Akt and NFκB activation, thereby reducing TNF-α production (Lee et al., 2006). Similarly, Choi and colleagues have documented that an A3 receptor agonist, LJ529, significantly reduces the release of IL-1β, TNF-α, and MCP-1 in LPS-treated microglial cells. However, they also observed that the IC₅₀ values of LJ529 for the inhibition of cytokine/chemokine release were much higher (in the micromolar range) than the nanomolar Ki values for the A3 receptor (Tchilibon et al., 2005). These observations suggest that the A3 receptor may not play a major role in cytokine release in microglia. Choi and colleagues further demonstrated that A3 receptor engagement inhibits microglial migration toward chemoattractants by downregulating the spatiotemporal expression of Rho GTPases (Choi et al., 2011). As a consequence, LJ529 suppresses microglia/monocyte accumulation into the ischemic brain and protects the brain in an in vivo model of cerebral ischemia. In contrast to this conclusion, a recent publication showed that A3 receptor signaling may promote the ADP-induced process extension and migration of microglia (Ohsawa et al., 2012).

Interactions between the A3 receptor and other P1 adenosine receptors have also been reported. The A3 receptor and A2A receptor recipically inhibit each other and thereby orchestrate cellular responses to stressors (van der Putten et al., 2009). Specifically, the A2A receptor mediates inhibitory signaling and the A3 receptor mediates activation in microglia. In unstimulated microglia, these two receptors counterbalance each other to maintain homeostatic equilibrium. However, under conditions of inflammatory stimulation, the decrease in A3 receptor-mediated signaling sensitizes microglia to A2A receptor-mediated inhibitory signaling. Thus, the A3 receptor may be a dynamically regulated suppressor of the A2A receptor.

In short, there is a dearth of conclusive knowledge on the function of A3 receptors. Further studies are required in order to clarify how this receptor affects microglial functions.

4.2. P2 ATP

P2 receptors are divided into five subclasses: P2X, P2Y, P2Z, P2U, and P2T. P2Y, P2U, and P2T purinergic receptors are metabotropic whereas P2X and P2Z receptors are ionotropic. Microglia express multiple types of these receptors, including P2X and P2Y (James and Butt, 2002). P2 receptors bind to ATP and UTP/UDP. Many CNS cells, including neurons and astrocytes, release ATP as a transmitter to mediate intercellular communication (Butt, 2011). During acute brain injuries such as ischemic stroke, neurons enter rapid energy failure, resulting in loss of ionic gradients and consequent depolarization. This is followed by intracellular ATP depletion and outflow of ATP from cells (Juranyi et al., 1999; Latini and Pedata, 2001). In addition to neurons, ATP is also released from vascular cells and blood cells (Bune et al., 2010). Furthermore, the release of UTP/UDP from damaged neurons has been observed after brain injury (Koizumi et al., 2007). These purines trigger various responses in microglia through corresponding P2 receptors. Interestingly, the impact of ATP may depend on the model and/or ATP concentration. Microglia exposed to 3 mM ATP acquire an amoeboid phenotype (Butchi et al., 2010) whereas exposure to 0.6 – 1 mM ATP has the opposite effect and induces the ramified phenotype (Wollmer et al., 2001). Microglia can also release ATP, suggesting the presence of an autocrine loop (Ferrari et al., 1997c).

4.2.1. P2X4

P2X4 is the most abundant P2X receptor in the CNS (Buell et al., 1996). Direct application of ATP or LPS increases P2X4 receptor expression in cultured microglia (Gong et al., 2009; Raouf et al., 2007). Functionally, microglial P2X4 receptors are involved in chemotaxis by the modulation of PI3K signaling (Ohsawa et al., 2007). An interaction between P2X4 and mu-opioid receptors has been shown to enhance microglial migration in response to morphine in a PI3K/Akt-dependent manner (Horvath and DeLeo, 2009). P2X4 also mediates the production of proinflammatory cytokines from microglia upon exposure to hypoxia (Li et al., 2011).

An induction of microglial P2X4 has been observed in CNS disease models that involve inflammation, such as SCI, cerebral ischemia, and preterm hypoxia-ischemia (Cavaliere et al., 2003; Li et al., 2011; Schwab et al., 2005; Tsuda et al., 2003; Ulmann et al., 2008; Wixey et al., 2009). Specifically, microglial P2X4 receptor expression is essential for spinal inflammatory pain processes (Schwab et al., 2005; Tsuda et al., 2003; Ulmann et al., 2008). Pharmacological blockade or molecular suppression of the P2X4 receptor leads to a reduction in tactile allodynia after SCI, whereas intraspinal administration of microglia in which P2X4 receptors has been pre-activated produces tactile allodynia in normal rats (Tsuda et al., 2003). Mechanistically, stimulation of the P2X4 receptor by ATP causes the SNARE-mediated release and synthesis of BDNF, which is dependent on extracellular Ca²⁺ influx and activation of p38-MAPK (Ulmann et al., 2008) (Figure 4). The elevation in BDNF in turn weakens the tonic inhibition of lamina I GABAergic interneurons (Coull et al., 2005) and results in the development of allodynia.

Figure 4. P2X4 receptors in tactile allodynia after spinal nerve injury.

Spinal nerve injury may induce P2X4 receptor upregulation in microglia through the activation of fibronectin/integrin, Lyn kinase, PI3K/Akt, and MEK/ERK-dependent signaling pathways. Chemokine CCL2 and CCL21 are also involved in transforming quiescent microglia into the P2X4-expressing phenotype through interaction with microglial CCR2, CXCR3, and CCR7 receptors. Stimulation of the P2X4 receptor by ATP causes the SNARE-mediated release and synthesis of BDNF, which is dependent on extracellular Ca²⁺ influx and activation of p38-MAPK. The elevated BDNF in turn weakens the tonic inhibition of lamina I GABAergic interneurons and results in the development of allodynia.

The molecular mechanisms mediating the upregulation of P2X4 receptors on microglia have been actively studied. Spinal nerve injury may induce P2X4 receptor upregulation and neuropathic pain through fibronectin/integrin, Lyn kinase, PI3K/Akt, and MEK/ERK-dependent mechanisms (Tsuda et al., 2008a; Tsuda et al., 2009b; Tsuda et al., 2008b) (Figure 4). Another study demonstrated that CCL21 is rapidly expressed in injured small-sized primary sensory neurons and transported to their central terminals in the dorsal horn after SCI. The upregulation of neuronal CCL21 enhances microglial P2X4 receptor expression and leads to tactile allodynia (Biber et al., 2011). Chemokine CCL2 and cytokine IFNγ have also been shown to transform quiescent microglia into the P2X4-expressing phenotype (Beggs et al., 2012; Toyomitsu et al., 2012; Tsuda et al., 2009a). Further elucidation of microglial P2X4 receptor upregulation and function following SCI may help identify new therapeutic targets for neuropathic pain.

4.2.2. P2X7

Microglial P2X7 receptor stimulation by exogenous ATP triggers a rapid and sustained transmembrane influx of extracellular Ca²⁺ as well as an efflux of intracellular K⁺ (Ferrari et al., 1996; Gudipaty et al., 2003; Kahlenberg and Dubyak, 2004). Such dramatic changes in intracellular ion homeostasis activate multiple signaling molecules, including NFκB, ERK1/2, p38, and JNK1/2 (Ferrari et al., 1997d; Parvathenani et al., 2003; Shiratori et al., 2010; Takenouchi et al., 2007). Several transcription factors whose activation are involved in the expression of inflammatory genes, such as NFκB, nuclear factor of activated T cells (NAFT), and cyclic AMP response element-binding (CREB), are also activated by P2X7 receptors in microglia (Ferrari et al., 1999; Kataoka et al., 2009; Shiratori et al., 2010). Consequently, P2X7 engagement induces the secretion of pro-inflammatory mediators such as TNF-α, IL-1β, superoxide, NO, CXCL2, and CCL3 in microglial cells (Gendron et al., 2003; Hide et al., 2000; Kataoka et al., 2009; Parvathenani et al., 2003; Shiratori et al., 2010). In particular, the P2X7 receptor has been shown to be critical for IL-1β production in microglial cells (Ferrari et al., 1997b; Takenouchi et al., 2009). LPS or other inflammatory stimulants induce the synthesis and cytoplasmic accumulation of microglial pro-IL-1β, which is biologically inactive. The P2X7 receptor, in response to relatively high concentrations of ATP, induces the rapid processing and massive release of mature IL-1β in LPS-primed microglia (Sanz and Di Virgilio, 2000). Thus, purinoreceptors mediate the release of pro-inflammatory cytokines from microglia under conditions of high ATP.

Interestingly, mechanically stimulated astrocytes also release ATP. Astrocyte-derived ATP induces the shedding of IL-1β-containing vesicles from nearby microglia in a P2X7-dependent manner, suggesting that astrocyte-microglia crosstalk through ATP/P2X7 interactions plays a critical role in the release of IL-1β from microglia (Bianco et al., 2005). Mechanistic studies further reveal that P2X7 receptor engagement activates IL-1β-converting enzyme/caspase 1 in microglia (Sanz and Di Virgilio, 2000; Takenouchi et al., 2008), which in turn induces actin filament reorganization and the shedding of IL-1β-containing vesicles (Takenouchi et al., 2008).

In addition to regulating inflammatory cytokine secretion, the P2X7 receptor might be important for microglial cell death and proliferation in a dose- or duration-dependent manner. Prolonged activation of the P2X7 receptor leads to the formation of membrane pores in microglia and results in the induction of apoptotic/necrotic cell death (Ferrari et al., 1997a; Ferrari et al., 1999; Takenouchi et al., 2008). In contrast, moderate P2X7 receptor activation demonstrates trophic effects and increases microglial proliferation, especially under pathological conditions associated with increased TNFα (Bianco et al., 2006; Monif et al., 2009).

Disease states appear to upregulate P2X7 expression on microglia. For example, microglial P2X7 receptors are upregulated in response to injuries such as cerebral ischemia (Franke et al., 2004). Such receptor upregulation enhances microglial activation and might be a therapeutic target for brain injury. For instance, P2X7 receptor antagonists have been shown to protect against transient global cerebral ischemia reperfusion injury by reducing inflammation (Chu et al., 2012). In addition, P2X7 receptor antagonists suppress seizures, confer neuroprotection during status epilepticus, and reduce neuropathic pain (Engel et al., 2012; He et al., 2012).

4.2.3. P2Y receptors

P2Y receptors are a family of G-protein coupled receptors that bind ATP, ADP, UTP, and UDP. The expression and functions of P2Y6 and P2Y12 in microglia have been studied. For example, the P2Y6 receptor is known to be important for microglial phagocytic activity (Koizumi et al., 2007). Koizumi and colleagues first reported that P2Y6 receptors are upregulated in microglia following kainic acid-induced neuronal injury and act as a sensor to detect the UTP/UDP signals released from damaged neurons. The stimulation of P2Y6 by UTP/UDP enhances microglial phagocytosis and therefore facilitates the removal of damaged cells and cellular debris. P2Y6 receptors also regulate other aspects of microglial activity. For example, an in vitro study with glial cell and slice cultures documented that UDP-P2Y6 interactions resulted in elevated expression of CCL3 and CCL2 in microglia and astrocytes, suggesting a possible role of the P2Y6 receptor in chemokine production from glial cells (Kim et al., 2011).

In contrast to the function of P2Y6 in microglial phagocytosis, P2Y12 has been shown to be important for microglial motility (Haynes et al., 2006). P2Y12 is expressed predominantly in microglia in the brain (Sasaki et al., 2003), but is dramatically reduced after microglial activation (Haynes et al., 2006; Orr et al., 2009). Microglia from P2Y12 deficient mice exhibited normal baseline motility but are unable to extend processes or migrate toward ATP/ADP in vitro or in vivo, resulting in reduced directional process extension toward sites of cortical damage (Haynes et al., 2006). Further studies confirm that the P2Y12 receptor is critical for microglial chemotaxis and have identified several signaling molecules downstream of P2Y12 activation, such as integrin β1, PLC, and PI3K/Akt (Irino et al., 2008; Ohsawa et al., 2007; Ohsawa et al., 2010)

In summary, purinergic receptors are involved in multiple microglial functions, including process motility, chemotaxis, phagocytosis, and cytokine/chemokine production. A close collaboration between these receptors is critical in mediating efficient neuron-microglia interactions. For instance, damaged neurons as well as other CNS cells release ATP and UTP. Whereas ATP serves as a ‘find me’ signal to activate P2Y12-mediated cell motility, UTP and its degradated product UDP serve as ‘eat-me’ signals to trigger P2Y6-mediated phagocytosis. It is also notable that the expression of purine receptors on microglia is dynamic and activation state-specific (Orr et al., 2009; van der Putten et al., 2009). For example, LPS-treated microglia exhibit marked A2A receptor upregulation along with a loss of P2Y12 receptors. Such changes switch microglial motility from chemoattraction to repulsion, suggesting that a context-dependent shift in purinergic receptor signaling may orchestrate microglial responses to stressors (Orr et al., 2009).

5. Phosphatidylserine receptors and bridging proteins

5.1 Phosphatidylserine receptors and bridging proteins in neuron-microglia interactions

Phosphatidylserine is normally expressed on the inner leaflet of the plasma membrane in healthy cells. However, phosphatidylserine exposure on the exoplasmic leaflet is a characteristic feature of cell apoptosis (Fadok et al., 1998). The recognition of phosphatidylserine by its receptors on phagocytes is believed to be an essential ‘eat me’ signal for the clearance of dying cells (Ravichandran, 2011). A variety of engulfment receptors on phagocytes are capable of binding to phosphatidylserine. These phosphatidylserine receptors function in two modes: direct recognition and bridged recognition. The receptors directly recognizing phosphatidylserine include the phosphatidylserine receptor (PSR) (Fadok et al., 2000), the T cell immunoglobulin and mucin (TIM) family members (Kobayashi et al., 2007; Miyanishi et al., 2007), Stabilin2 (Park et al., 2008) and brain angiogenesis inhibitor I (BAI-I) (Park et al., 2007). For indirect phosphatidylserine recognition, bridging proteins such as milk fat globule-EGF factor 8 (MFG-E8) and growth arrest specific gene 6 (Gas6) are essential (Hanayama et al., 2002; Leonardi-Essmann et al., 2005). These proteins usually have two binding sites; one site binds to the phosphatidylserine and the other binds to receptors on phagocytes. Together, these sites mediate phosphatidylserine-dependent phagocytosis of apoptotic cells.

Phosphatidylserine receptors are critical for the microglial clearance of apoptotic neurons (Hirt and Leist, 2003; Witting et al., 2000). As one might expect, O-phospho-L-serine, a molecule that mimics the phosphatidylserine head group and blocks microglial phosphatidylserine receptors, interferes with the uptake of apoptotic neurons by microglia (Witting et al., 2000). Specifically, the expression of PSR has been detected on quiescent microglia and is enhanced by LPS stimulation (De et al., 2002). The engagement of PSR may promote the phagocytosis of apoptotic neurons and shift microglial activation toward an anti-inflammatory phenotype (De et al., 2002; Hirt and Leist, 2003).

Bridging proteins are also important for microglial function. MFG-E8 is expressed in microglia and secreted in response to soluble factors, such as CX3CL1, which is released from degenerating neurons (Fuller and Van Eldik, 2008; Leonardi-Essmann et al., 2005; Li et al., 2012; Noda et al., 2011) (Figure 5). MFG-E8 can bind to phosphatidylserine on apoptotic neurons and simultaneously engage integrin αvβ3 on microglia. Elevated MFG-E8 expression therefore results in accelerated microglial clearance activity. Moreover, MFG-E8 enhances production of heme oxygenase-1 (HO-1) and reduces neurodegeneration.

Figure 5. Phosphatidylserine receptors and bridging proteins are critical for the microglial clearance of apoptotic neurons.

MFG-E8 is expressed in microglia and secreted in response to soluble factors, such as CX3CL1, which is released from degenerating neurons. MFG-E8 can bind to phosphatidylserine on apoptotic neurons and simultaneously engage integrin αvβ3 on microglia. Elevated MFG-E8 expression therefore results in accelerated microglial clearance activity. Moreover, MFG-E8 enhances production of heme oxygenase-1 (HO-1) through Nrf2 activation. Another bridging protein, Gas6, bridges phosphatidylserine residues on the surface of apoptotic cells to the Axl/Mer family of tyrosine kinases (Mer, Axl, and Tyro3) on microglia and stimulates microglial phagocytosis of apoptotic neurons in a vav/Rac-dependent manner. Gas6 receptors mediate an anti-inflammatory response to apoptotic cells by inhibiting NF-κB activation.

The bridging protein Gas6 has also been detected on microglia (Grommes et al., 2008) (Figure 5). Gas6 bridges phosphatidylserine residues on the surface of apoptotic cells to the Axl/Mer family of tyrosine kinases (Mer, Axl, and Tyro3) on microglia and stimulates microglial phagocytosis of apoptotic neurons in a vav/Rac-dependent manner. Similar to MFG-E8, Gas6 receptors mediate an anti-inflammatory response to apoptotic cells in addition to boosting phagocytosis.

A question that has undergone years of debate is whether phosphatidylserine exposure alone is sufficient to induce phagocytosis. Although early studies claim that phosphatidylserine by itself could send a signal strong enough to elicit phagocytic responses in macrophages, subsequent evidence challenged this conclusion by showing that many cells are not taken up by phagocytes despite the presence of phosphatidylserine on their outer leaflet (Borisenko et al., 2003; Suzuki et al., 2010). A second signal has therefore been proposed that may work in concert with phosphatidylserine to send an efficient ‘eat-me’ signal to phagocytes (Gardai et al., 2005). Other possible explanations include the requirement of a threshold number of exposed phosphatidylserine molecules (Borisenko et al., 2003) and the preference for modified phosphatidylserine (Tyurin et al., 2008) for phagocytosis induction. However, all these hypotheses still remain to be tested.

5.2 Phosphatidylserine receptors and bridging proteins in CNS injuries

As mentioned above, bridging proteins such as MFG-E8 and Gas6 mediate indirect phosphatidylserine recognition. A recent study demonstrates neuroprotective and anti-inflammatory effects of recombinant MFG-E8 protein in a mouse model of stroke, indicating the potentially therapeutic value of this protein in neurological disorders (Cheyuo et al., 2012). Gas6 deficiency exacerbates microglial activation as well as oligodendrocyte loss after cuprizone-induced demyelination (Binder et al., 2008). Consistent with its role as a microglial tyrosine kinase that is bridged to phosphatidylserine residues on apoptotic cells, Axl deficiency reduces microglia/macrophage recruitment and impairs myelin debris clearance following demyelination during EAE (Weinger et al., 2011). In contrast, treatment with exogenous Gas6 ameliorates microglial activation and enhances remyelination following cuprizone-induced demyelination (Binder et al., 2011; Tsiperson et al., 2010). Interestingly, Gas6 not only modulates microglial activity, but also independently influences the oligodendrocyte response after CNS damage.

6. Pattern recognition receptors (PRR)

It is well known that microglial activation can be mediated by a superfamily of PRR that recognize molecular motifs comprised of pathogen associated molecular patterns (PAMPs) and danger associated molecular patterns (DAMPs). PAMPs are a set of molecular determinants derived from invading pathogenic bacteria, fungi, or viruses that command immediate neutralization and subsequent removal by the innate immune system. DAMPs, on the other hand, are derived from cellular debris, intracellular proteins/enzymes released from necrosis following catastrophic events such as ischemic injury and spinal cord injury, as well as neurological disorders such as MS, AD, and PD.

Different classes of PRR such as Toll like receptors, carbohydrate binding receptors, Scavenger receptors, Rig like receptors (Rig like helicases) (Lech et al., 2010), Nod like receptors (Rosenzweig et al., 2011) and Aim2 like receptors (Brunette et al., 2012) with distinct recognition motifs and subcellular localizations work in concert to neutralize exogenous pathogens and toxic material. Among them, Toll like receptors, carbohydrate binding receptors, and scavenger receptors have all been implicated in CNS injuries. The function of PRR in microglial activation has been extensively reviewed elsewhere. We will only discuss them briefly here.

6.1. Toll like receptors (TLRs)

Toll like receptors are mammalian homologues of the Toll gene identified in Drosophila nearly three decades ago. In mammals, 13 isoforms designated as TLR1 to TLR13 have been identified. Only 11 of them are expressed in human cells (Hanke and Kielian, 2011) and 9 in microglia (Bsibsi et al., 2002; Lee and Lee, 2002). In the brain, TLR signaling is a part of the innate immune system. TLRs undergo homodimerization to achieve specific ligand or agonist recognition, with the exception of TLR2, which associates with TLR1 and TLR6 in order to recognize bacterial lipopeptides (Kawai and Akira, 2007).

Induction of several TLRs has been strongly associated with activation of microglial cells in response to inflammation (Lehnardt et al., 2007; Olson and Miller, 2004; Trinchieri and Sher, 2007). Among those TLRs, TLR2 and 4 have been shown to modulate the extent of the inflammatory response and neuronal demise. Immediately after stroke, neuronal degeneration-induced inflammatory responses lead to the production and release of pro-inflammatory cytokines and excitotoxins via activation of TLR2 and TLR4 in microglia (Caso et al., 2007; Hua et al., 2007; Kilic et al., 2008; Lehnardt et al., 2007). Indeed, stroke-induced injury, infarct size and neurological deficits are all significantly reduced in mice deficient in TLR2 and TLR4 (Tang et al., 2007). Inhibition of neuronal TLR2 and TLR4 prevents JNK activation in ischemia (Tang et al., 2007). In addition, TLR4 deficiency decreases TNFα, iNOS and COX2 production (Pradillo et al., 2009), potentially resulting in greater tolerance towards ischemic injury. In contrast, TLR3 or TLR9 activation has no impact on stroke outcome.

6.2. Carbohydrate binding receptors

Recent reports suggest that carbohydrate structures attached onto proteins during post-translational modification can interact with microglia to modulate their activation status and function. Understanding the relationship between carbohydrate moieties and carbohydrate binding receptors on microglia may provide a novel means of mitigating inflammation-induced damage in various neurological disorders.

6.2.1. Galectins

In mammals, there are 15 isoforms of galectins that are further subdivided into galectin1 and galectin3 based on sequence homologies and functions (Liu, 2002). Typically, galectins contain a highly conserved motif known as the carbohydrate recognition domain (CRD) that enables their direct association with carbohydrates, specifically, β-galactosides. It has been speculated that galectins recognize galactose-containing oligosaccharides on a group of glycoproteins and glycolipids in vivo (Liu, 2002). Microglial galectin1 and galectin3 are functional counterparts in modulating inflammatory responses. The former facilitates anti-inflammatory events and the latter promotes pro-inflammatory cascades (Rabinovich and Toscano, 2009). Therefore, a tightly temporally titrated expression of specific subsets of galectins on microglia helps ensure a balanced and productive inflammatory response.

An increase in galectin3 has been correlated with microglial activation in ischemic injury (Lalancette-Hebert et al., 2012; Rotshenker, 2003), leading to the demise of hippocampal neurons (Satoh et al., 2011). Indeed, preconditioning strategies or hypothermia both result in the downregulation of microglial galectin3 and attenuate these deleterious events (Lalancette-Hebert et al., 2012; Satoh et al., 2011). Alternatively, upregulation of microglial galectin1 has been shown to suppress pro-inflammatory responses in MS (Starossom et al., 2012), supporting the counter functions of galectin1 and galectin3 that were originally proposed by Rabinovich and Toscano (Rabinovich and Toscano, 2009). In addition, supplementation with recombinant galectin1 enhances astrocytic BDNF production and reduced neuronal apoptosis in ischemic injury (Qu et al., 2010). These responses suggest an inherently protective function of microglial galectin1 and a potentially toxic role for galectin3. In non-neuronal systems, galectins may engage multiple signaling pathways such as Akt, p38, ERK, NF βB, and JNK (Chung et al., 2012; Koh et al., 2008; Pineda et al., 2012). The impact of galectin on signal transduction in the brain has yet to be defined.

6.2.2. Selectins

Selectins are part of a receptor family that associates with sialyated and fucosylated glycan structures (Varki, 2007). Among them, E- and P-selectins are expressed in endothelium and platelets cells, respectively. In contrast, exclusive expression of L-selectins has been reported in astrocytes and microglia (Jeon et al., 2008). Selectins are involved in cell-cell interactions during the initiation of inflammatory responses (Varki, 2007). P- and L-selectins are bound by sulfatides, a class of sulfated glycosylceramides. Thus, sulfatide-induced inflammatory responses in the brain may be mediated via microglial selectins. The concomitant reduction of pro-inflammatory cytokines by microglial L-selectins also suggests that these carbohydrate-binding receptors modulate neuroinflammation (Jeon et al., 2008).

Our understanding of the roles of selectins in various neurological disorders lies in its infancy. In stroke, E- and P-selectin have been shown to affect platelet and endothelial activities (Beer et al., 2011; Jaremo et al., 2012), which in turn modulate inflammatory responses and degeneration (Ma et al., 2012). Interestingly, polymorphisms in L-selectin appear to increase the risk of stroke (Wei et al., 2011). Nevertheless, whether microglial selectins are mechanistically involved in this disease still remains elusive because epidemiological studies are correlational in nature.

6.3. CD36

Scavenger receptors are a set of PRR that have been used to identify modified lipoproteins liberated during cellular damage and degeneration. Scavenger receptors facilitate the removal of these lipoproteins. To date, over 20 structurally distinct scavenger receptors with overlapping ligands have been identified (Husemann et al., 2002). Nevertheless, the exact structural recognition motifs for all scavenger receptors have not been elucidated. Scavenger receptors can be classified into 6 categories (A, B, C, D, E and F). Among them, CD36, which belongs to SCARB, is closedly related to CNS injuries.

CD36 participates in HDL-mediated transport of lipids and cholesterol to the liver (Krieger, 1999). This transport process, termed “reverse cholesterol transport,” is the principal mechanism by which HDL prevents the cholesterol buildup that leads to atherosclerosis (Trigatti et al., 1999). In addition, CD36 is found on macrophages and dendritic cells derived from monocytes and may mediate the uptake of degenerating neutrophils, lymphocytes, and fibroblasts (Albert et al., 1998; Puente Navazo et al., 1996; Ren and Savill, 1995; Savill et al., 1992). CD36 is highly expressed in mouse neonatal microglia but is expressed at lower levels in adult microglia (Coraci et al., 2002). The function of microglial CD36 is not yet clear. However, it is known that CD36 protein expression is increased in the ischemic brain, predominantly in microglia/macrophages. CD36-null mice exhibit reduced infarct volumes after MCAO with attenuated ROS production in the early stages after reperfusion. These findings support a role for CD36 in ischemia-induced oxidative stress and brain injury. Despite these toxic roles for CD36, CD36 has also been shown to promote hematoma absorption via microglial phagocytosis, thereby protecting brain cells from intracerebral hemorrhage-induced damage (Zhao et al., 2009; Zhao et al., 2007). A recent study further reveals that CD36 deficiency worsens short-term outcomes from focal stroke in neonates due to enhanced inflammation and diminished removal of apoptotic neuronal debris (Woo et al., 2012). These studies suggest that the function of CD36 in brain injuries depends on the nature of the injury, the stage of damage, or the age of the victims.

7. Future directions

In summary, a broad range of surface receptors are expressed on microglia and mediate microglial recognition of ‘On’ or ‘Off’ signals on other host cells as well as invading microorganisms. Integrated actions of these receptors result in tightly-regulated biological functions, including cell mobility, phagocytosis, inflammatory mediator release, and the induction of acquired immunity. Over the last few years, significant advances have been made towards deciphering the signaling mechanisms related to these receptors and identifying their contribution to specific cellular functions. As described above, a recurring theme in the mechanism of action of microglial receptors is the orchestration of opposing actions. The juxtaposition of ITAM against ITIM, SHP-1 against SHP-2, galectin1 against galectin3, and numerous other examples reveal an elegant means of fine-tuning microglial responses in response to different challenges so that homeostasis of the CNS can be achieved. Also as a result of these juxtrapositions, microglial sensitivity to different stimuli has an unexpectedly high degree of resolution. The complexity of microglial responses to environmental challenges is also reflected in their dual-faced nature, where they can switch from a protective to a toxic phenotype in ways that are not completely understood. Indeed, there remain many unresolved challenges in this field, some of which will be discussed below.

First and foremost, there remain serious technical obstacles in studies of microglia. For example, the in vivo studies of microglial receptors have been largely dependent on pharmacological inhibition and/or genetic deletion of the receptor of interest. However, most microglial receptors are also expressed on other types of cells, posing a potential confound to the interpretation of pharmacological or gene deletion studies. In other words, the effects of global inhibition or ablation of these receptors may also reflect cells other than microglia. Second, the characterization of microglial responses is usually based on immunostaining or biochemical methods using microglia-specific markers. However, the lack of an antibody specific to either microglia or macrophages precludes our ability to distinguish between local microglia and circulating macrophages that were recruited to the brain following injuries. Third, because microglia react to even minute changes in brain milieu, it is critical to examine these cells in their natural environment as far as possible. Recently, novel techniques such as microglia-specific conditional knockout, transgenic cell labeling, in vivo microglia imaging/tracking have been developed and optimized to help overcome some of these difficulties (Sieger and Peri, 2013; Xiao et al., 2011). However, new quantitative imaging probes are still necessary to permit researchers to monitor signaling events within living cells. In combination with live cell imaging of microglia, such a technique would greatly improve our understanding of microglia-specific signaling dynamics in different microenvironments (Sieger and Peri, 2013).

Besides technical challenges, another obstacle to studying microglia is the relative lack of information on the interactions between different microglial receptors and their combined downstream effects. It is known that multiple distinct receptors on microglia are activated simultaneously or sequentially in response to the same pathological challenge. Despite our knowledge of individual receptors, the interplay between these receptors remains uncharted territory. To address this issue, we need to describe the spatiotemporal kinetics and mode of action for each specific receptor in isolation and in combination with other receptors, the sequence of receptor activation and deactivation, and, finally, how the coordinated actions of multiple receptors fulfill the subtle biological roles of microglia. Recent research has made some progress toward this goal. For example, accumulating evidence suggests that multiple microglia receptors, including CD36, CD47, integrinα6β1, TLR4, TLR2 and scavenger receptor A, are co-activated in response to Aβ and form a large receptor complex to mediate microglial phagocytosis of Aβ fibrils and the subsequent activation of pro-inflammatory intracellular signaling pathways (Bamberger et al., 2003; Reed-Geaghan et al., 2009; Wilkinson et al., 2006). In contrast to such cooperative functions, FcγR activation has been shown to stimulate inhibitory signaling through another microglial receptor, signal regulatory protein α (SIRPα), which in turn inhibits FcγR- and complementary receptor-mediated phagocytosis (Gresham et al., 2000; Oldenborg et al., 2001). Again, advances in real time imaging technologies may offer a better experimental platform to further clarify our understanding of how different surface receptors work together in both time and space in order to modulate microglial functions in CNS injuries.

It must also be borne in mind that the expression and function of many microglial receptors are sexually dimorphic and age-dependent. For example, the expression of several ‘Off’ receptors, including CX3CR1 and CD200, decreases with age (Lyons et al., 2007; Wynne et al., 2010), which might contribute to the age-related dysregulation of microglial responses to CNS injuries (Wong, 2013). In addition, the expression of several P2 nucleotide receptors on microglia has been shown to vary with age and gender (Crain et al., 2009). For example, the microglial expression of P2X4 is age-dependent and significantly lower in females than in males. Furthermore, the expression of P2X7 or P2Y6 is low in neonates but significantly increases from 21 days of age onwards, with no obvious gender difference. Many CNS diseases are more common in one gender and increase in occurance in elderly populations. Thus, elucidating how these gender-specific and age-related differences influence the function of microglia will provide valuable insights on the contribution of these receptors to the aging process and their roles in CNS diseases.

Many experimental strategies that target microglial receptors have been used for the treatment of neurological disorders in animal models (Binder et al., 2011; Clark et al., 2007; Pertwee, 2009; Tsiperson et al., 2010; Zhuang et al., 2007). Translating this research into the clinic will be far more challenging because we do not yet have a comprehensive view of microglial receptors. For example, more information on the topographic distribution of receptors on non-microglial cells under normal and pathological conditions would be useful to predict possible off-target effects. Furthermore, better strategies to selectively target microglia will be essential to achieve drug specificity (Cucchiarini et al., 2003; Minami et al., 2012). The temporal kinetics of microglial behavior will be particularly critical in deciding the treatment regimen. As one example, inhibition of CXCL12-CXCR4 signaling in early phases of stroke may restrain microglial activity. However, interference with microglial activity must be avoided in advanced stages when neurovascular remodeling becomes active. Finally, researchers must also bear in mind that the long-term modulation of microglial activities may negatively affect homeostasis and immune defense and result in undesired side effects such as opportunistic brain infections and the buildup of necrotic cellular debris. Fortunately, the pace of research on microglia has increased exponentially in the past decade, not to mention since del-Hortega seminal observations of microglia/macrophages almost a hundred years ago. If we continue to advance our understanding of microglial receptors at this more recent pace, new strategies to modulate microglial activity without adversely impacting immune status may lie on the horizon.

Highlights.

• Microglia is the first line of immune defense in the central nervous system

• Integrated actions of microglial receptors result in tightly regulated biological functions.

• Microglial homeostasis is carefully maintained by multiple counterbalanced strategies, including ‘On’ and ‘Off’ receptor signaling.

• Understanding of the signaling mechanisms related to these receptors and identifying their contributions to specific cellular functions will advance our knowledge of many CNS injuries.

Glossary

Aβ

amyloidβ

AD

Alzheimer’s disease

AGEs

advanced glycation endproducts

COX2

cyclo-oxygenase-2

CNS

central nervous system

DAP12

DNAX-activating protein of molecular mass 12 kilodaltons

Dok

downstream of tyrosine kinase

ERK

Extracellular signal-regulated kinases

Gas6

growth arrest specific gene 6

HMGB1

high-mobility group box 1

HO-1

heme oxygenase-1

IFN-γ

interferon-gamma

IgSF

immunoglobulin superfamily

IL

interleukin

JNK

c-Jun-N-terminal protein kinase

iNOS

inducible nitric oxide synthase

ITAM

immunoreceptor tyrosine-based activation motif

ITIM

immunoreceptor tyrosine-based inhibition motif

LPS

lipopolysaccharide

MAPK

mitogen-activated protein kinase

m-csf

macrophage colony-stimulating factor

MFG-E8

milk fat globule-EGF factor 8

MMP

matrix metalloproteinase

MCP-1

monocyte chemotactic protein-1

MIP

macrophage inflammatory protein

MS

multiple sclerosis

NFκB

nuclear factor-κB

NO

nitric oxide

PD

Parkinson’s disease

PI3K

phosphoinositide-3-kinase

PKC

protein kinase C

PLCγ

phospholipase Cγ

RAGE

Receptor for advanced glycation endproducts

PSR

phosphatidylserine receptor

RANTES

regulated on activation, normal T cell expressed and secreted

ROS

reactive oxygen species

SCI

spinal cord injury

SHIP

SH2 domain-containing inositol polyphosphate 5′ phosphatase

SHP1

SH2 domain-containing phosphatase-1

TBI

traumatic brain injury

TGF-β

transforming growth factor-β

TLR

toll-like receptor

TNFα

tumor necrosis factor α

TREM

Triggering receptors expressed on myeloid cells