Most evidence for microglial cytokine production and effects derives from studies employing pure or mixed primary cultures. However, in vitro studies may miss features that depend on other cell types or conditions only provided by a tissue environment. Regional properties are blended when cultures are prepared from larger anatomical structures. Developmental steps reached by the time of culture preparation influence microglial behavior in vitro (Draheim et al., 1999). Cultures are routinely prepared from perinatal tissue, a period when microglia is not yet at resting state. This may contribute to the alerted status of cultured microglia. Cell lines are very useful when larger amounts of material are required, but bear the risk of deviated properties, including failure to produce the full spectrum of inducible factors (Stohwasser et al., 2000). Brain slice preparations are used for (electro)physiological and imaging-based recordings, but come with the unavoidable limitation of acute tissue damage (Brockhaus et al., 1993; Boucsein et al., 2000). Rapid reactions (e.g., phosphorylation events) are probably initiated before microglial cells are examined. Acute slices are thus not always suitable. Even though preparation-induced cytokines would only appear after hours, experimental stimulations would necessarily top on the mechanical stress. As a compromise, organotypic cultures allow microglial cells to return to a more resting status, as shown by morphology, proliferation markers, and whole-tissue release activity (Mertsch et al., 2001).
Constitutive vs. Inducible Synthesis
Cytokines are mostly produced upon “request.” Constitutive synthesis by resting microglia is thus unlikely, except for neurotrophins (Fig. 1) (Elkabes et al., 1996). Lack of (e.g., immunocytochemical) evidence for expression may still not rule out production and simultaneous release of undetectable but functional amounts (Hanisch, 2001b). Basal cytokine mRNA (IL-1) levels can be elevated in vitro, but the degree of preactivation is below the threshold required for measurable release of protein, the actual carrier of function. Nevertheless, membrane-associated versions (e.g., of TNFα) can be detected in cell-associated form, and several cyto- and chemokine receptors are constitutively expressed (Fig. 2). Stimulation rapidly induces cyto/chemokine gene transcription and effective release while intracellular protein amounts remain mostly very low. Members of the IL-1 family may exist as preformed precursors, allowing instant conversion and release of mature cytokine upon stimulation (Fassbender et al., 2000). Little is known about the time frames for desensitization and over which a microglial cell can maintain release activity without exhaustion.
Figure 1. Simplified scheme illustrating some major routes of communication between microglia and other resident as well as invading cells as they involve cytokines/chemokines. Healthy neurons may inform microglia about their normal activity via constitutively calming inputs (e.g., fractalkine) or release of activity-associated (co)transmitters, such as ATP. Resting microglia or microglia at low degree of activation (alerted microglia) could support neuronal function and survival by production of neurotrophic factors. Upon transformation due to a challenge by dangerous signals, activated microglia may not only chemoattract leukocytes via chemokine synthesis, but also produce cytokines and other factors (NO, reactive oxygen intermediates) with potential toxicity for neurons and activating or aggravating potential for astrocyte involvement and further microglial recruitment. Signals for microglial activation may also derive from severed or irreversibly damaged neurons by, for example, disruption of calming inputs (disturbed fractalkine signaling) or massive release of ATP. Lymphocytes and other leukocytes may send feedbacks to microglia to adjust the profile of invasion-supporting signals. Astrocytes will have a major input on microglia (e.g., by production of M-CSF or IL-6, not shown) but also on neurons by release of substances that may even provide harm, such as glutamate.
Download figure to PowerPoint
Figure 2. Expression of cyto- and chemokine receptors in microglia. Stimulation by environmental signals will cause the transformation of the resting microglia to an activated state. Those signals can derive, for example, from a confrontation with infectious agents. Bacterial lipopolysaccharide (LPS) is often used as a tool to simulate a challenge by gram-negative bacteria and to study the microglial activation process. As a consequence, microglial cells change morphological features and functional properties, including the induction of cyto- and chemokine release, altered ion channel and cell surface antigen expression, chemotactic movement, and phagocytotic behavior. Certain cyto- and chemokines can themselves trigger some of these events and have a strong modulatory influence on the outcome of a microglial challenge. Cyto- and chemokines act via specific receptors or receptor complexes, their associated cytosolic effectors, such as G-proteins and kinases, and the recruitment of multiple intracellular signaling pathways.
Download figure to PowerPoint
Factors Triggering and Modulating Microglial Cytokine Induction
Events triggering cyto/chemokine production are numerous, but only few signals and their receptive structures are molecularly identified (Hanisch, 2001a). Viral envelopes, bacterial cell wall components, and other infectious agents (prion protein) cause macrophage/microglia activation (Heppner et al., 2001). Lipopolysaccharides (LPS) of gram-negative bacteria serve as standard agents in mimicking infections. Proteoglycans and (lipo)teichoic acid as structural determinants of gram-positive strains are powerful cytokine inducers (Draheim et al., 1999; Prinz et al., 1999; Hanisch et al., 2001). Bacterial DNA may have own contributions. Inflammatory mediators, such as platelet-activating factor (PAF), lipids, or complement factors are similarly effective. Certain viruses even confuse the defense system by encoding for cyto/chemokine-like structures, receptor homologues, or cytokine-binding protein.
CNS injury liberates signals that instruct microglial transformation. Some may originate from the cytosol or the membranes of disintegrating or stressed cells. Trauma signals can be subtle and affect (micro)glia in the vicinity of neuronal somata while the primary insult is set farther away (Streit et al., 2000a). Microglial activation might be initiated or modified by molecules commonly (co)released in neurotransmission (Fields and Stevens, 2000; Honda et al., 2001). Monitored by microglia, low levels of ATP and its degradation products would indicate normal neuronal activity (Fig. 1). They even counteract cytokine production as induced by an activating stimulus, “convincing” microglia of still intact cell functions and thus preventing drastic responses (data not shown). High ATP levels, as they facilitate microglial reactions, would signal excessive cellular activity or damage (Hide et al., 2000; Sanz and Di Virgilio, 2000). Extracellular [K+] exceeding normal ranges could have a similar meaning (Fig. 1).
Injury may release factors (including cytokines) that are bound to the extracellular matrix in a functionally “silent pool”, but carry latent microglia-activating signal character. Microglia could also sense inundating serum proteins that are normally denied CNS entry (e.g., proteases, their inhibitor complexes, lipid-loaded albumin) (Coughlin, 2000; Gingrich and Traynelis, 2000; Möller et al., 2000). Serum factors can also fulfill functions as accessory agents in microglial responses. Examples are LPS-binding protein (LBP) and serum components required for full microglial activation by gram-positive bacteria (Prinz et al., 1999).
Proteins with disease-related production, processing, and aggregation, such as amyloid β (Aβ) in Alzheimer disease (AD), can reportedly stimulate microglia, including its release properties. In conjunction with other stimuli, Aβ aggeregates seem to irritate microglial cells chronically as they concentrate around AD plaques. Clusters of activated microglia then produce factors (such as IL-1, TNFα) that can drive neurotoxic cascades that in turn recruit more microglia (Fig. 1).
Other factors could trigger microglial activation by a kind of off-signaling. They may usually maintain the resting status. Interruption of their calming input can then result in microglial activation. The neuronal/endothelial/lymphoid glycoprotein CD200 constitutively provides an inhibitory influence on macrophages/microglia (Hoek et al., 2000; Wright et al., 2000). There are also candidates among the cyto/chemokines (Fig. 1).
Correlation of Cytokine Induction With Other Functions of Activated Microglia
Morphology, surface antigen expression, electrophysiological properties, and the migratory, phagocytotic, as well as release behavior undergo dramatic changes when microglia is challenged (Streit et al., 2000a; Hanisch, 2001a). But are all these changes always part of a stereotypic response profile? Are some “executive limbs” induced in parallel, whereas others follow a sequence? Cytokine release can be inhibited by blocking signaling pathways without affecting other activation parameters (Prinz et al., 1999; Hanisch et al., 2001). The shape of microglia can considerably vary and differ from the resting phenotype and still the cells do not produce cytokines. Nevertheless, some mechanistic links exist, e.g., between K+currents and the ability to release IL-1–like cytokines (Ferrari et al., 1997). Initial release of certain cytokines, such as TNFα, has an (autocrine) impact on further release activities. Similarly, cytokines affect the expression of cell surface molecules as a prerequisite for chemotactic movements and cellular interactions, including antigen presentation or induction of apoptosis. Cytokine systems of pivotal importance for microglia are discussed below.
IL-1/IL-18 Family Members
IL-1 is a crucial microglial effector cytokine. As an immunostimulatory/proinflammatory signal, IL-1 has a strategic position in innate defense and immune responses (Dinarello, 1998a, 1998b). Cells of lymphoid and myeloid lineage are main sources, IL-1α being mostly cell-associated, IL-1β being the major soluble form. The numerous targets include T- and B-cells, monocytes, and macrophages/microglia. IL-1 receptor antagonist (IL-1ra), the third family member, serves in the control of IL-1 activities.
IL-1 seems to be implicated in many neuropathological scenarios (Allan and Rothwell, 2001). Activated glial cells, namely microglia, and invading immune cells can serve as major sources upon infection, ischemia, stroke, excitotoxicity, and mechanic injury (Hanisch, 2001a). Following neurotrauma, IL-1 is rapidly released, even faster than expected for de novo synthesis (Fassbender et al., 2000). Even though the etiology of CNS disorders varies, IL-1 appears to be a common link in processes leading to neuronal death (Rothwell et al., 1997; Loddick et al., 1998; Rothwell and Luheshi, 2000). Suppression of endogenous IL-1 has thus often neuroprotective effects.
Apart from inflammation, IL-1 has probably functions in cell proliferation and differentiation during CNS development (Giulian et al., 1988a). It can modulate synaptic efficacy in neuronal populations (Dunn et al., 1999; Vitkovic et al., 2000), especially of the hippocampus (Katsuki et al., 1990; Schneider et al., 1998; Luk et al., 1999; O'Connor and Coogan, 1999). IL-1R accessory protein-like molecule (IL1RAPL), a novel IL-1R–related protein (Carrie et al., 1999), is enriched in the hippocampal formation. Mutations are found in X-linked mental retardation and link a defect in an IL-1R–like molecule directly to disturbed CNS function. IL-1 influences neuro-endocrine-immune circuits by enhancing the activity of hypothalamic neurons and pituitary cells (Berkenbosch et al., 1987, 1989; Besedovsky and del Rey, 1996; McCann et al., 1997; Tringali et al., 1998). Fever induction, soporific effects, suppression of food intake, and behavioral changes are thought to be part of an IL-1–regulated sickness behavior (Konsman et al., 2002). This may involve transport-assisted BBB penetration of circulating IL-1, effects triggered at this border, true central IL-1 actions, and afferent fiber pathways stimulated by peripheral IL-1 (Banks et al., 1991; Banks and Kastin, 1997). Indirect influences on regulatory nuclei are mediated by IL-1–inducible mediators, especially by IL-6 and an array of releasing factors. Part of the IL-1–induced CNS effects may serve in normal control mechanisms (Alheim and Bartfai, 1998; Maier et al., 1998; Cartmell et al., 1999; Luheshi et al., 1999; Takahashi et al., 1999). Even though these functions are still poorly understood (Hopkins and Rothwell, 1995), modulation of neuronal and glial properties might belong to the pleiotropic activity spectrum. Their dysregulation upon pathologically increased IL-1 levels would add to homeostatic disturbances.
IL-18, identified as IFNγ-inducing factor (IGIF), is an IL-1–like cytokine by structure and certain activities in innate and adaptive immunity (Okamura et al., 1995; Bazan et al., 1996; Kohno and Kurimoto, 1998; Dinarello, 1999; Akira, 2000). Mainly produced by macrophages/monocytes, IL-18 affects T-, B-, and NK cells. Synergistic IFNγ induction by IL-12 and IL-18 drives Th1 responses. IL-1 and IL-18 share recruitment of cytosolic adapters and kinases, which translates into overlapping activation of downstream elements and transcription factors (Hanisch, 2001a). IL-18 and IL-18R signaling was also demonstrated in CNS cells and tissues (Conti et al., 1997, 1999; Culhane et al., 1998; Prinz and Hanisch, 1999; Wheeler et al., 2000). Astrocytes and microglia appear to be a source. Localization to the hypothalamus, pituitary, and adrenal cortex suggests neuroendocrine involvement, whereas IL-18 mRNA in the cortex, striatum, hippocampus, and cerebellum indicate additional physiological roles (Culhane et al., 1998). The developmental course of IL-18 expression points to some role in brain ontogeny (Prinz and Hanisch, 1999). Effects on immune cells and the strong induction of IFNγ and other factors (TNFα) suggest that IL-18 is critical for driving neuroimmune/inflammatory circuits in infection and multiple sclerosis (Wildbaum et al., 1998; Fassbender et al., 1999). Increased CNS levels may also result in cell death without obvious participation of immune cells (Rothwell et al., 1997; Loddick et al., 1998).
Emerging concepts suggest that IL-1, IL-18, and their receptor/effector systems belong to ancient superfamilies participating in cell communication and defense mechanisms throughout the phylogenetic tree (Hanisch, 2001a). The family of IL-1/IL-18–like structures just recently expanded by five “family of IL-1” (FIL1) or “IL-1 homologue” (IL-1H) molecules (Kumar et al., 2000; Smith et al., 2000).
IL-2 is known as the T-cell growth factor. It is produced mainly by T-cells and drives T- and B-cell proliferation and differentiation. IL-15 was discovered for its ability to substitute for some IL-2 activities, but is produced by a much broader variety of cells. IL-2 and IL-15 share the same receptor subunits for signaling, and their complexes containing IL-2 receptor α (IL-2Rα), IL-2Rβ, IL-2Rγ, and IL-15Rα are widely expressed throughout brain tissues and ontogeny (Hanisch et al., 1997a). Numerous, often very potent CNS effects have been reported for IL-2, supported by molecular and biochemical evidence for the presence of the receptors (Hanisch and Quirion, 1996; Hanisch, 2001b). Whether CNS effects are in vivo preferentially mediated by IL-2 or IL-15 is still uncertain. Microglial cells express all four receptor molecules and respond to IL-2 and IL-15 with cytosolic events and increased survival in culture. In combination with LPS or IFNγ, IL-2 can enhance microglial NO production while IL-15 has rather an attenuating influence (Hanisch and Quirion, 1996; Hanisch et al., 1997a). This illustrates that the two cytokines can have different effects on the same cell type. Microglia produces IL-15, as shown by mRNA and protein (Hanisch et al., 1997a; Prinz et al., 1998). However, little is known about the microglial IL-2/IL-15 system in CNS pathologies. IL-2 is toxic at elevated CNS tissue levels and has a plethora of neuroregulatory and neuroendocrine consequences, suggesting that microglial IL-2/IL-15 activity could have multiple influences on the CNS, including the pituitary (Hanisch and Quirion, 1996; Hanisch, 2001a, 2001b).
IL-4, -10, -13, and TGFβ share features of anti-inflammatory, immunosuppressive and neuroprotective actions. Much of these outcomes can be attributed to a downregulation of (micro)glial production of cytokines, e.g., IL-1 and TNFα, or the attenuation of their secondary release effects (Chao et al., 1993; Dinarello, 1997a; Kitamura et al., 2000; Pahan et al., 2000). IL-4 and IL-13 also interfere with IL-1 bioactivity by enhancing IL-1ra synthesis (Dinarello, 1997b, 1997c). In addition, these cytokines can alter the microglial cell surface molecule expression (Chao et al., 1993; Suzumura et al., 1994; Raivich et al., 1999a; Sawada et al., 1999; Wei and Jonakait, 1999; Szczepanik et al., 2001). Grouping of these factors just for their neuroprotective outcomes should not ignore more specific functions or individual modes of actions, as known from their effects on B-cells, mast cells, and thymocytes.
TGFβ, a factor with multiple biological activities in many cells and playing roles in various tissue developments, immune responses, and wound healing, reduces proinflammatory cytokine and chemokine production. Most notably, TGFβ1 was recently shown to reduce AD plaque load in an animal model of AD (Wyss-Coray et al., 2001). The assumed mechanism of Aβ clearance involves the phagocytotic activity of microglia. TGFβ might either promote microglial Aβ phagocytosis directly or via increased levels of Aβ-binding proteins, but the key message of the observation is that the plaque burden could be lowered by boosting the activity of an endogenous clearance mechanism.
IL-6 can be produced by T-, B-, bone marrow, and endothelial cells, macrophages, and various CNS cell types, including microglia (Lee et al., 1993). Targets are T-, B-, plasma, bone marrow, and liver cells. Microglia also expresses IL-6R. Like IL-1 and TNFα, IL-6 is considered a proinflammatory cytokine acting in the initiation and coordination of inflammatory responses and limiting the spread of infectious agents. IL-6 particularly helps to initiate and regulate acute-phase responses, a complex of adjustments in metabolic and executive organ functions (liver, immune cells) and circulating serum components that assist in host defense. CNS activities relate to fever induction, sleep, increased pain perception, reduced food intake, and neuroendocrine mobilization of energy stores. At cellular level, IL-6 may therefore mediate many of the physiological effects known for IL-1 as it is induced by IL-1. Interestingly, IL-6 can have both pro- and anti-inflammatory outcomes (Campbell, 1998; Raivich et al., 1999a). The tendency for one over the other net effect is likely determined by the simultaneous presence of other factors (cytokines). Microglia seems to provide IL-6 especially in early phases of CNS insults (Raivich et al., 1999a). Subsequently, IL-6 may act on astrocytes to involve these cells in the orchestration of attempts for tissue repair (Raivich et al., 1999a). IL-6 and M-CSF are thought to be major mediators of microglial activation (Raivich et al., 1996; Streit et al., 2000b).
The type I interferons, IFNα, IFNβ, IFNω, and IFNτ, show similarities in structure and receptor binding, whereas IFNγ (type II) is structurally distinct and has its own receptor (Hanisch, 2001b). IFNα/β play pivotal roles in the resistance of mammalian cells against viral infection, while additional functions relate to immune cells (Akbar et al., 2000; Hanisch, 2001b). IFNγ, immune interferon, is mostly triggered by antigenic stimulation and T-cell activation. CD4+ and CD8+ T-cells and natural killer (NK) cells serve as major sources. IFNγ has a critical role in the decision of whether Th1 or Th2 immune responses are favored. Th2 cells produce only small IFNγ amounts, but are targets of an antiproliferative cytokine effect. Macrophages have been found to synthesize IFNγ upon costimulation with IL-12 and IL-18. IFNγ induces major histocompatibility (MHC) class I and II complexes—even in cells normally not expressing these surface structures—and cell adhesion molecules, hence supporting antigen presentation, induction of humoral and cell-mediated immune responses, and interactions of lymphocytes with the vascular endothelium. Induction of IL-1, TNFα, and chemokines, together with multiple other cellular effects (IFNγ controls more than 200 genes), suggests a pivotal role in the regulation of immune activity (Hanisch, 2001b).
Functional studies revealed IFN receptors on glial and neuronal cells (Hanisch, 2001b). Presence of IFNs or their receptors in normal tissues points to various CNS functions. IFNα/β appear in the CNS especially upon infection, elevated levels of IFNγ during pathological conditions. They correlate with T-cell invasion, but endogenous expression has been demonstrated as well. Interestingly, an IFNγ-like substance has been frequently seen in numerous neuronal cells (Hanisch, 2001b).
In mircroglia, IFNγ causes induction and upregulation of many cell surface molecules, namely MHC class I and II, intercellular adhesion molecule I (ICAM-I), immune-accessory molecules B7 (CD80/86), leukocyte function-associated molecule 1 (LFA-1), LPS receptor (CD14), Fc and complement receptors, changes in the proteasome composition, as well as release of cytokines (TNFα, IL-1, IL-6), complement (C1q, C2, C3, C4), and NO. It also has profound modulatory effects on the cytokine induction of various microglia-stimulating agents (Stohwasser et al., 2000; Hanisch, 2001b). IFNβ and TGFβ can antagonize some of these cellular actions. IFNγ effects on the cytotoxic, phagocytic, and antigen-presenting features of microglia thus resemble those known for (other) macrophages (Raivich et al., 1999a; Hanisch, Raivich et al., 2001b).
Recent findings emphasize that the cyto/chemokine production by microglia (and astrocytes) is under a strong, complex, and context-determined control of IFNγ (data not shown). Chemokines release modulation in glia could contribute to the IFNγ shaping of immune cell invasions (Tran et al., 2000). In addition to its microglia-activating effects, IFNγ can also induce apoptosis through simultaneous upregulation of Fas and FasL (Badie et al., 2000).
TNF and Relatives
The prototypic proinflammatory cytokine TNFα is a product of monocytes/macrophages, dendritic cells, or lymphocytes and has activities in virtually all cells, including the regulation of growth and differentiation. TNFα production in the CNS can be attributed to neurons, astrocytes, as well as microglia (Lee et al., 1993; Cheng et al., 1994; Zhang and Tracey, 1998; Prinz et al., 1999). Glial cells are often involved as both donors and recipients of cytokine signals. Autocrine loops may serve to maintain an activated status, but could also aggravate destructive activities under pathological conditions. TNFα could play such a role (Bezzi et al., 2001).
Increased levels of TNFα in the brain have been observed especially after injury or ischemia, in bacterial and viral infections, as well as in multiple sclerosis and AD. On injury, activated microglia is an early and prominent source of TNFα (Gabay et al., 1997). Depending on the stimulus, even a short exposure can be very effective and result in strong and lasting release response (Hanisch et al., 2001). Thus, in CNS insults, rapid induction of microglial TNFα production could critically influence subsequent events. Microglia can also release soluble (s)TNF-R II, hence affect the bioavailable TNFα pool and probably regulate potential TNFα consequences, but TNFα:sTNF-R ratios vary with the type of stimulation.
Harmful outcomes of increased TNFα relate to its activity to promote inflammation and edema, although TNFα administration in animals has been reported to cause only minor inflammatory changes until paired with a pathogenic stimulus (Angstwurm et al., 1998). High CSF levels indicate poor prognosis in infection. Microglial TNFα also seems to be critical in boosting a recently identified chemokine-induced and TNFα-mediated glutamate release from astrocytes (Fig. 1) (Bezzi et al., 2001). In addition, TNFα has direct toxic effects on neuronal structures and myelin. In contrast, TNFα is shown to promote neural cell survival and proliferation. Deficiency in TNF signaling makes animals more susceptible to conditions evoking neuronal damage, probably relating to an impaired NGF induction. Apparently, low levels of TNFα have neuroprotective consequences, whereas high levels can cause destructive outcomes. TNFα thus stands for the occasionally opposing actions of several cytokines (Carlson et al., 1999; Mason et al., 2001).
TNFα and TNF-R belong to families of ligands and related receptors, including TNFα, lymphotoxins (LTs), Fas ligand (FasL), CD40 ligand (CD40L), and TRAIL as ligands and TNF-R I and II, LTR, Fas (CD95), CD40, TRAIL-R1, and TRAIL-R2, as well as p75NTR (NGF, neurotrophin receptors) as the receptive structures (Ashkenazi and Dixit, 1998, 1999; Ware et al., 1998). Activation of these receptors by their respective ligand causes multiple and heterogeneous effects, ranging from apoptosis to enhanced proliferation. These ligand-receptor systems are crucial for cell selection and programmed cell death, also in neural development and synaptogenesis, and participate in microglia-mediated cell communication and toxicity. Special attention was recently attributed to the CD40L-CD40 system in microglial activation by Aβ (Tan et al., 1999).
Chemokines are chemotactic cytokines acting through G-protein–coupled receptors, which currently undergo an enormous expansion in terms of identified structures and functions (Locati and Murphy, 1999; Rossi and Zlotnik, 2000). Extravasation and tissue trafficking of leukocytes depend on the timed and coordinated expression of chemoattractants (Asensio and Campbell, 1999). Chemokine roles in cell migration are assisted by immunoregulatory functions for Th1/Th2 responses and the control of cytokine profiles. Expression in CNS cells and tissues has mainly been shown in pathologies and their animal models (Asensio and Campbell, 1999; Glabinski and Ransohoff, 1999; Locati and Murphy, 1999; Mennicken et al., 1999; Zhang et al., 2000). Mechanical injury, neuroinflammatory processes, brain tumors, demyelinating diseases such as multiple sclerosis, viral and bacterial infections, stroke, and neurodegenerative processes such as AD all seem to contain chemokine actions via constitutive and upregulated chemokine receptors.
Chemokine receptors occur on neuronal and glial populations (Asensio and Campbell, 1999; Glabinski and Ransohoff, 1999). Neurons express an array of CCR and CXCR. Regional differences suggest specific functions. While expression patterns under normal conditions suggest neuromodulatory or supportive chemokine effects, receptors on glial cells could serve in migratory events upon pathological changes. IL-8R, CXCR2, CXCR3, CXCR4, CCR3, CCR5, and CX3CR1 are among the receptors reported for microglia in vitro and/or in vivo (Glabinski and Ransohoff, 1999; Mennicken et al., 1999; Biber et al., 2001). Microglia can produce GROα (KC), MIP-1α, MIP-1β, MIP-2, MCP-1, RANTES, IP-10, and IL-8 in response to experimental stimulation by bacterial agents, Aβ peptides, as well as cytokines, such as TNFα and IL-1. Even though the sets are not strictly complementary, the profiles of receptor and inducible chemokine expression suggest that activated microglia could serve in further microglial recruitment. Via chemokines, microglia can also affect (even support) neurons and astroglia. Besides meningeal (and perivascular) macrophages and ependymal cells, activated micro- and astroglia could conceivably contribute to leukocyte attraction and guidance (Nau and Bruck, 2002).
Cytokines such as IFNγ can dramatically modify the patterns of micro- and astroglial chemokine production (data not shown). Individual chemokines can show little change, whereas the induced release of others becomes potently increased or suppressed. The regulatory IFNγ influence is not only selective and varies with the nature of the microglia-activating stimulus, but reveals a coordinated change in the composition of the chemokine “cocktail” as it would preferentially attract cell types. Indeed, the cellular composition of leukocytic infiltrates is known to undergo changes during neuroinflammation. IFNγ, primarily released by invading T-lymphocytes, could represent an important signal to instruct (micro)glial cells on how to modify their chemokine profile as the neuroinflammatory process proceeds. Leukocyte recruitment is a necessary element. However, it may occasionally have a deleterious net consequence, since release products of invading cells can be quite toxic. Therefore, interference with leukocyte invasion has often beneficial outcomes on the course of CNS infections (Nau and Bruck, 2002).
While certain chemokine receptors show promiscuous behavior in accepting various ligands, others are more selective. The SDF-1-CXCR4 (stroma cell-derived factor 1 and its receptor) system appears to be important for CNS development and could be rather relevant for microglia and its interaction with other CNS cells. Recent findings suggest that CXCR4 stimulation induces release of TNFα in astro-microglial communication (Bezzi et al., 2001). Moreover, certain ligand-receptor systems appear to be crucial for CNS responses to injury, e.g., secondary lymphoid tissue chemokine (SLC)/CXCR3 (Biber et al., 2001). The fractalkine-CX3CR pair has special implications for microglia. Disturbances of its signaling may be sufficient to trigger or enhance microglial activation (Fig. 1). Fractalkine, occurring in soluble and membrane-bound forms, is predominantly found on neurons, whereas CX3CR1 mainly associates with microglia (Harrison et al., 1998; Nishiyori et al., 1998; Maciejewski-Lenoir et al., 1999). Neuronal fractalkine may not only support microglial survival (Boehme et al., 2000), it may organize a constitutive “calming” effect on microglia (Zujovic et al., 2000). Fractalkine message is reduced on neuronal damage and the molecular forms change as CX3CR1-expressing microglia accumulates in the affected region (Harrison et al., 1998). TNFα and IL-1 can upregulate fractalkine in astrocytes, and (soluble) fractalkine can induce microglial migration (Maciejewski-Lenoir et al., 1999). Nevertheless, several neurons can also express CX3CR1 coupled to Akt-mediated survival signaling (Meucci et al., 2000). Moreover, microglia may itself produce fractalkine (Zujovic et al., 2000), allowing for autocrine signaling. When added during microglial stimulation, it attenuated TNFα release while its neutralization drastically augmented the cytokine output, raising the hypothesis that the chemokine may play a role in containing microglial activation.
The repertoire of cytokines that are either produced or sensed by microglia is even broader (Raivich et al., 1999a, 1999b). Colony-stimulating factors, such as M-CSF or GM-CSF, and IL-3 have also control on microglial properties, including proliferation. Microglia may partially serve itself for supply (Gebicke-Haerter et al., 1994). M-CSF has been discussed as a major factor in glial activation (Raivich et al., 1996; Schwaiger et al., 1998; Streit et al., 2000a). IL-12, a cytokine produced, for example, upon infection, can be synthesized by microglia (Taoufik et al., 2001). The cells produce especially large amounts of the inducible IL-12p40 peptide, which is shared by heterodimeric IL-12p70 as well as IL-23 and, as a homodimer, can antagonize IL-12p70 (Oppmann et al., 2000; Hanisch et al., 2001). Production of IL-12p70 (consisting of p35 and p40) appears to depend on cosignaling of cytokines such as IFNγ. Via production of IL-12p40 and/or IL-12p70, microglia could modulate the development of Th1 vs. Th2 immune responses in the CNS (O'Garra and Arai, 2000). On the other hand, these cells are also subject to IL-12p70 and p40 effects, as illustrated by a modulation of their cyto/chemokines and NO production (Aloisi et al., 1997; Stalder et al., 1997; Suzumura et al., 1998; Pahan et al., 2001; Prinz et al., 2001).
Effects of Microglial Cytokines on Astrocytes, Oligodendrocytes, and the Endothelium
Astrocytes and endothelial cells produce various cytokines upon appropriate stimulation. By virtue of cytokines, however, activated microglia seems to orchestrate many (subsequent) astrocytic responses and endothelial features. Some microglial release products may primarily carry a toxic potential for oligodendrocytes.
Astrocytes serve the metabolic homeostasis, contribute to the BBB formation, and engage with a reestablishment of tissue integrity following injury. Hypertrophy (astrogliosis) with enhanced expression of cyll-type–specific glial fibrillary acidic protein is a common response to nervous tissue injury (Eddleston and Mucke, 1993). Astrogliosis and glial scar formation are needed to close wounds, to wall off necrotic areas, and to stabilize regions of neural injury. Astrocytes further assist in protective and repair mechanisms by secretion of neurotrophic factors, removal of neurotoxins, and elimination of excitatory neurotransmitters (Eddleston and Mucke, 1993; Giulian, 1993). While astrocytes support microglial functions, such as phagocytosis, activated microglia/macrophages and their soluble products are obviously required for the evolution of astrocytic responses (Balasingam et al., 1996; Smith and Hoerner, 2000). Cytokines have profound effects on astrocytes and can drive astrocytic proliferation. IFNγ, IL-1, IL-2, IL-6, TNFα, and M-CSF were shown to associate with astrogliosis (Hanisch, 2001a). Especially IL-1 is needed, and IL-1ra is sufficient to prevent astroglial proliferation (Giulian et al., 1988a, 1988b, 1994a, 1994b). On the other hand, IL-10 can effectively attenuate astrogliosis, likely due to a downregulation of microglial cytokines (Balasingam and Yong, 1996).
Oligodendrocytes can respond to IL-1, -2, -6, and IFNγ with reduced or enhanced growth (depending on the experimental condition) and may occasionally express certain cytokines (Hanisch, 2001a). Several cytokines may play harmful roles in demyelinating diseases, such as multiple sclerosis (Antel et al., 1996; Merrill and Benveniste, 1996; Merrill and Scolding, 1999; Martino et al., 2000). While some cytokine involvement is indirect by assisting in immune mechanisms or by stimulating (micro)glial production of reactive oxygen and nitrogen species, TNFα can be directly toxic to oligodendrocytes, damage myelinated structures, and prevent remyelination (Hanisch, 2001a).
Cytokines of microglial production can potentially induce the release of endothelial mediators and stimulate the expression of cell surface molecules, namely adhesion molecules, required for leukocyte extravasation. Transport and exclusion of serum molecules as regulated by the BBB are likely affected when microglia becomes focally activated. Several cyto- and chemokines with angiogenetic (and angiostatic) properties are also known release products of activated microglia. Via proteases, phagocytosis, and the production of wound healing-associated cytokines (including chemokines and growth factors), microglia has a pivotal role in tissue repair.
Microglial Cytokines Affecting Neuronal Activity
Cytokines have multiple demonstrated effects on various neuronal populations. There are reports on (out)growth- or differentiation-promoting and growth factor deprivation-counteracting activities for populations of cortical, hippocampal, septal, striatal, and pituitary neurons as developmental contributions of cytokines (including growth factors themselves). Even though the physiological involvement is still largely enigmatic, cytokines can obviously also modulate neurotransmitter release via constitutively expressed receptor complexes (Hanisch, 2001a, 2001b). Often, these cells or anatomical structures could be shown by various techniques to contain the respective receptors. The hippocampus and the hypothalamus are the major—but not the only—brain structures for those reported cytokine actions. Release of acetylcholine, noradrenaline, dopamine, serotonin, and GABA has been shown to change upon treatment of cells or ex vivo tissue preparations with cytokines, such as IL-1, IL-2, interferons, and TNFα (Rothwell and Hopkins, 1995; Hanisch, 2001a, 2001b). Most often, cytokines affected evoked transmitter release while not changing basal levels. Biphasic modes of action and region-specific influences were reported (Hanisch et al., 1993; Seto et al., 1997). Stereotactic cytokine injections of minute amounts into brain structures clearly cause neurophysiological reactions and even behavioral alterations, e.g., changes in membrane potential, discharge frequencies, short- and long-term potentiation, EEG activity, afferent sensory transmission, sleep, locomotion, exploratory activity, or memory (Vitkovic et al., 2000; Hanisch, 2001b). Interestingly, disruption of cytokine signaling can also interfere even with higher CNS activities (Petitto et al., 1999). Cytokines characteristically synthesized and/or sensed by microglia could therefore influence the function of neighboring neurons. Microglia may start to disturb CNS functions when massively releasing these cytokines. Finally, some proliferation-inhibiting, apoptosis-inducing, and toxic effects could turn microglia into an intrinsic source of potentially harmful factors.
Contribution of Microglial Cytokines to Neuro-Immune-Endocrine Interactions
Induction of sickness behavior with fever and food intake suppression or the adjustment of endocrine activities are effects attributed to cytokines in physiological adaptations to immune system challenges or stressful situations (Hanisch, 2001a; Konsman et al., 2002). Several cytokines, including IL-1, IL-6, and TNF, are known to influence the release of hormones from the pituitary or the adrenal gland directly and to cause potent (mainly activating) effects on endocrine axis via hypothalamic nuclei. Similarly, homeostatic functions relating to temperature control and food intake have been shown to be sensitive targets of central cytokine actions (Zhang and Tracey, 1998; Vitkovic et al., 2000). Neuronal circuits and soluble messengers are structural substrates for a communication between the immune, nervous, and endocrine systems. Here it should be stressed that glial cells, namely microglia, could contribute to this communication via cytokine release and that other mediators, such as NO, could also influence hypothalamic neurons and neurosecretory cells in a microglia-to-neuron signaling.
There is thus sufficient evidence also to consider receptor-mediated (functional) disturbances due to inappropriate cytokine synthesis by microglia, although most harmful outcomes may result from direct toxicity and the (auto)destructive potential of dysregulated host defense.