The concept of the brain as an ‘immune privileged’ organ has been modified significantly with the realization that interactions between the nervous and immune systems are important in many aspects of disease and injury. Peripheral infection, inflammation and injury activate afferent neuronal and homoral signals to the brain, which in turn influence or regulate specific aspects of the host acute phase response. Many molecules and pathways associated with immune or inflammatory responses have been identified within the central nervous system and are activated in response to peripheral disease or injury and diverse neurological disorders. Cytokines are polypeptide mediators most widely associated with the inflammation and immunity in the periphery and play a key role in many of these interactions. This review will focus mainly on the ‘proinflammatory cytokines’, interleukin-1 (IL-1), IL-6 and tumour necrosis factor α (TNFα), which are expressed by, and act on the brain, and have been implicated in many aspects of neu-roimmunology. The host defence response, also known as the acute phase response, includes changes on both local (eg inflammation within specific tissues) and systemic functions. The central nervous system regulates important aspects of this response including fever, anorexia, hypermetabolism, neuroendocrine changes (most notably activation of the hypothalamic-pituitary-adrenal, axis), alterations in behaviour, cardiovascular and reproductive function and even some aspects of immune activation. Much of our understanding of the local and afferent pathways regulating these changes has derived from studies on fever. Fever is still one of the most widely used diagnostic tools in clinical medicine, it is a phylogen-etically old response which has even been observed in poi-kilotherms, can be readily and continuously monitored in free-moving subjects and animals by remote radiotelemetry, and appears to share common mechanisms with several central nervous system-dependent acute phase responses (Kluger, 1991; Roberts, 1991). Studies on fever and other neuroimmune responses in rodents have most often used systemic (intraperitoneal or intravenous) administration of bacterial lipopolysaccharide as a stimulus, which has provided much valuable information on fever and other central nervous system responses to infection (eg see Kluger, 1991). However, this approach has several potential disadvantages. High doses of lipopolysaccharide are required to induce fever in rats and mice, so injected lipopolysaccharide could act at numerous sites including in the brain itself, and it is not possible to study local pathways or mediators within specific sites of infection or inflammation. Nevertheless, lipopolysaccharide-induced fever is associated with dramatic increases in circulating IL-6 and release of cytokines in the brain and is inhibited by blocking IL-1 actions at both sites (Le May et al., 1990; Rothwell et al., 1991). We have previously demonstrated that a sterile abscess induced by an intramuscular injection of turpentine (which does not diffuse into circulation) also elicits marked fever in the rat and, like lipopolysaccharide, is associated with increases in circulating IL-6 of many orders of magnitude (Turnbull et al., 1992; Cooper et al., 1994; Miller et al., 1997b). Indeed circulating IL-6 levels correlate with the rising phase of fever, although there may be a threshold level for IL-6, or existence of co-factors which are required to act synergistically with IL-6 in pyrogenesis (Turnbull, 1993). However, in these and many other experimental systems, little or no increase in circulating IL-1 is detected in circulation, although blocking IL-1 by injection of a neutralizing antibody or recombinant interleukin-1 receptor antagonist (IL-1ra) significantly attenuates fever (Long et al., 1990; Miller et al., 1997a). These data question the hypothesis that IL-1 is the primary circulating endogenous pyrogen, but nevertheless indicate an important role in fever and suggest that it may act locally within infected or inflamed tissue. In order to address these issues we have studied the effects of administering pyrogens into a sterile, subcutaneous air pouch which allows sampling of local cytokine production and modification of cytokine activities within the air pouch without direct effects of the stimulus on distant tissues. Injection of lipopolysaccharide (100 μg kg−1) into such an air pouch in the rat elicits marked fever, of comparable magnitude to that seen in response to intraperitoneal injection of lipopolysaccharide. No increases in bioactive IL-1 or TNFα are identified in circulation (Figure 1) in response to either stimulus, but the bioactivity of both of these cytokines increases dramatically (30–1000 fold) in fluid sampled from the air pouch after lipopolysaccharide (Miller, 1996; Miller et al., 1997a, b). TNFα is the first cytokine detected (by 30 min after lipopolysaccharide) in the air pouch, followed by significant increases in IL-1 and IL-6. In agreement with earlier data on related models of systemic infection or inflammation, circulating IL-6 is increased by over 60 fold in response to air pouch administration of lipopolysaccharide and the time course is consistent with its direct involvement in fever. Increased IL-6, but not IL-1 or TNFα bioactivity, is also detected in cerebrospinal fluid within 2 h of lipopolysaccharide administration (Miller, 1996). It is likely that at least some of the IL-6 found in circulation and in the cerebrospinal fluid derives from the air pouch itself. We have observed that human recombinant IL-6 injected into an air pouch (in the absence of lipopolysaccharide) can subsequently be measured in circulation and cerebrospinal fluid in significant quantities by an immunoassay which does not detect rat IL-6 (Miller, Rothwell, Pitcher & Luheshi, unpublished observations) and IL-6 is transported across the blood brain barrier (Banks et al., 1994; Luheshi et al., 1994). Although IL-1 and TNF levels are not increased in cerebrospinal fluid, bioactivity of these cytokines is increased in push/pull perfusates from the hypothalamus of rats treated with lipopolysaccharide into an air pouch (Miller, Luheshi & Rothwell, unpublished data) or intraperitoneally (Klir et al., 1993). Cartoon depicting effect of injection of lipopolysaccharide (LPS) systemically (intraperitoneally) or into a subcutaneous air pouch injection on fever in the rat. Bacterial lipopolysaccharide (Δ), LPS, 100 μg kg−1 or vehicle (○) was injected after intraperitoneally (i.p.) or into a subcutaneous air pouch formed 6 days previously. Body temperature was recorded by remote, radiotelemetry in free-moving rats. Points shown are means and vertical lines indicate s.e.mean, n = 8. Although IL-1 is probably not an important circulating endogenous pyrogen, it does nevertheless appear to participate directly in the development of fever. Injection of IL-1ra, a recombinant form of the naturally occurring IL-1 receptor antagonist, into the subcutaneous air pouch almost completely abolishes the increases in core temperature and circulating IL-6 in response to lipopolysaccharide (LPS, Miller et al., 1997a). Intraperitoneal injection of IL-1ra causes more modest inhibition of fever, which may be due to entry of the antagonist into the central nervous system, since intracerebroventricular (i.c.v.) injection of a 20 fold lower dose of IL-1ra also suppresses the febrile response to air pouch LPS, without affecting circulating IL-6 production (Miller et al., 1997a), and IL-1ra can enter the brain by an active transport mechanism (Gutierrez et al., 1994). These data suggest that IL-1, TNFα and IL-6 are all produced locally within tissues in response to inflammation. IL-6 can be released into circulation and activate the brain either directly (via active transport into the central nervous system, Banks et al., 1994) or through release of other mediators. IL-1, TNFα and IL-6 have all been detected in and can be produced by cells in the brain (mainly glia) in response to systemic stimuli (Klir et al., 1993; Hopkins & Rothwell, 1995). These two cytokines probably interact to induce an increase in set point for body temperature via release of prostaglandins in the preoptic arterior hypothalamus. The additional role of neural afferents in the activation of fever or other responses to a peripheral inflammation, pyrogens or cytokines has been proposed. We have shown that injection of capsaicin (to cause C-fibre deafferentiation) in the rat attenuates the early phase of the febrile response to turpentine, but not lipopolysaccharide (Turnbull et al., 1992). Several groups have shown that subdiaphragmatic vagotomy inhibits fever and behavioural responses to lipopolysaccharide or IL-1 in the rat, indicating that the vagus is an important pathway for afferent neuroimmune signals (eg Bluthe et al., 1994; Watkins et al., 1994). However, there is some inconsistency between data from different laboratories and we have observed that although subdiaphragmatic vagotomy abolishes lipopolysaccharide or IL-1 induced behavioural changes in rats, it does not affect fever in the same animals (Luheshi, Bluthe & Dantzer, unpublished data). The evidence that cytokines act directly in the brain to elicit biological responses such as fever is several fold. Firstly, injection, i.c.v. or into the brain parenchyma, of IL-1α, IL-1β IL-6 or TNFα elicits maximal fever at doses several orders of magnitude lower than systemic pyrogenic doses. Secondly, inhibition of the actions of endogenous IL-1, IL-6 or TNF in the brain (by intracerebroventricular injection of IL-1ra or anti cytokine antibodies) reduces febrile responses to systemic pyrogenic stimuli (Klir et al., 1993; Cooper et al., 1994; Luheshi et al., 1997). The primary site of action of the pyrogenic cytokines in the brain is presumed to be the preoptic arterior hypothalamus and the fever induced by all of these cytokines is reduced or prevented by co-administration of cyclo-oxygenase inhibitors (2) (see Kluger, 1991). However, the nature and location of brain IL-1 receptors which mediate such responses remains enigmatic. Two IL-1 receptors have been identified. All known actions of the two IL-1 ligands (IL-1α and IL-1β) have been ascribed to interaction with the Type I (IL-1 RI, 80 kDa) receptor, while the Type II IL-1RII (68 kDa) receptor is believed to act as a ‘decoy’, which is shed from the membrane and does not signal IL-1 actions (Sims et al., 1993). mRNA for the IL-1 RI has been identified in rodent brain, predominantly in the dentate gyrus with little, if any, message in the hypothalamus (Takao et al., 1990; 1992; 1993). Furthermore, radio-iodinated IL-1α or β show very little binding in rat brain and some specific binding in the mouse brain, but again not in the hypothalamus (Takao et al., 1990; 1992; 1993; Marquette et al., 1995). These seemingly discordant findings may reflect methodological problems and the relative insensitivity of radioligand binding, particularly since IL-1 can elicit biological responses in cells with as few as ten receptors. The failure to detect mRNA for IL-1RI in the hypothalamus may also reflect poor sensitivity of in situ hybridization and/or the existence of additional atypical or novel IL-1 receptors. An accessory protein has recently been identified which forms a dimer with the IL-1RI (Greenfeder et al., 1995). This accessory protein has been identified in the brain (including the hypothalamus), but does not co-localize with IL-1RI in all brain regions (Liu et al., 1996). Using biotinylated IL-1ra, which may be more sensitive than binding of radioiodinated IL-1 and retains full biological activity, we have detected specific ‘binding sites’ in the rat hypothalamus. The binding of IL-1ra, which appears to be predominantly on neurones, is fully displaced by excess IL-1ra, IL-1α or IL-1β, indicating that it may reflect interaction with IL-1‘receptors’ (Luheshi, Rothwell & Toulmond, unpublished data). It is not yet known if these are the IL-1RI or other novel receptors. Several pieces of functional data are not consistent with the proposal that IL-1RI mediates IL-1 actions in the brain on fever. The dose-response profiles for development of fever in the rat are identical for recombinant rat IL-1α and β when these cytokines are injected systemically in the rat. However, when the same molecules are injected into the brain, IL-β is significantly more potent (Rothwell & Hopkins, 1995 and Figure 2). Furthermore, the actions of IL-1β in the brain on fever are blocked by co-administration (intracerebroventricular) of either a monoclonal antibody (ALVA42) raised to the IL-1RII (Luheshi et al., 1993), a corticotrophin releasing factor (CRF) receptor antagonist (Busbridge et al., 1989) or recombinant lipocortin-1 (Carey et al., 1990), while responses to IL-1α are unaffected by any of these treatments. Comparison of effects of injection of various doses of recombinant rat interleukin-1α (IL-1α) (•) and IL-1β (▪) either (b) intraperitoneally (i.p.) or (a) intracerbroventricularly (i.c.v.) on body temperature measured by remote radiotelemetry in the rat. Data are presented as area under the curve (AUC) for the temperature profile over the two hour period immediately after injection of IL-1. Points shown are means and vertical lines indicate s.e.mean, n = 8–10. These data suggest that IL-1α and IL-1β may act on different brain receptors and through different mechanisms to induce fever, possibly via novel and as yet unidentified receptors (for summary of actions of cytokines on fever see Figure 3). Figure indicating the role of proinflammatory cytokines within sites of infection in tissues, in circulation and in the brain, on the development of fever. It is well established that a number of cytokines are potent activators of the axis, which can act at the level of the hypothalamus, to induce expression and release of CRF, and at the pituitary release adrenocorticotrophic hormone (ACTH). In addition IL-1 can act through both of these mechanisms, and in some cases via direct actions in the adrenal, to stimulate glucocorticoid release (see Tkurnbull & Rivier, 1995; Buckingham, 1996). Glucocorticoids feed back as potent inhibitors of cytokine expression and action, not only on neuroendocrine responses, but also on fever (Coelho et al., 1995). The antipyretic effects of glucocorticoids are probably due, at least in part, to inhibition of prostaglandin synthesis and we have previously proposed that the peptide lipocortin-1 mediates suppressive effects of glucocorticoids on fever in the brain and periphery (Carey et al., 1990; Strijbos et al., 1993), which is consistent with the proposed actions of this peptide on the hypothalamic-pituitary adrenal axis (Buckingham, 1996). Glucocorticoids may also influence fever by suppression of the synthesis and/or release of CRF. The actions of some cytokines (IL-1β, IL-6 and IL-8) in the brain are dependent on CRF. Interestingly fever induced by IL-8 is not affected by cyclo-oxygenase inhibitors, but is reduced by dexamethasone (Coelho et al., 1995). Genetically obese rodents (ob/ob mice and fa/fa Zucker rat) show marked alterations in hypothalamic-pituitary adrenal activity and their obesity is significantly attenuated by adrenalectomy, glucocorticoid receptor antagonists or i.c.v. infusion of CRF (see Rothwell, 1990). These mutants exhibit significantly impaired febrile responses to i.c.v. injection of IL-1β (Figure 4), but respond normally to IL-1α, and the reduced responses to IL-1β can be restored by adrenalectomy (Carnie et al., 1989; Busbridge et al., 1990). Recently, the obesity of the ob/ob mouse has been ascribed to a mutation in the ‘ob gene’ which results in a failure to produce its product, the peptide leptin (Zhang et al., 1994), while fatty Zucker rats and dlb/dlb mice have a mutation in the leptin receptor (Caro et al., 1996; Chen et al., 1996; Tartaglia et al., 1996). Leptin, produced by adipose tissue, is believed to regulate energy balance in normal mammals by acting on the hypothalamus to control appetite and energy expenditure (Frederich et al., 1995; Caro et al., 1996; Tartaglia et al., 1996). Several pieces of evidence suggest a possible relationship between leptin, cytokines and neuroimmune interactions. Effects of central (i.c.v.) injection of recombinant mouse IL-1α and IL-1β on body temperature measured by remote radiotelemetry in (a) genetically obese, ob/ob mice and (b) fatty (fa/fa) Zucker rats. Data are presented as area under the curve over the two hour period immediately after injection of recombinant cytokine. Vehicle has no effect. Means ± s.e.mean are shown, n = 8. ***P < 0.001 vs lean. The leptin receptor is a member of the Class I cytokine receptor family (including IL-6) (Tartaglia et al., 1996), suggesting that it may share actions with these cytokines. Indeed we find that leptin is a very potent pyrogen which, when injected i.c.v. in rats, causes marked increases in body temperature at doses which parallel those which induce hypophagia (Luheshi, & Rothwell, unpublished data). Furthermore, leptin expression is induced by lipopolysaccharide, IL-1 or TNF et al., 1996), and leptin increases expression of CRF in the hypothalamus et al., 1996). leptin may influence energy balance and body temperature in response to systemic infection and injury and, play a key with in neuroimmune interactions and the effects of such as of and In addition to the role of cytokines in neuroimmune interactions between peripheral tissues and the brain, these molecules also participate in acute and disease and injury within the central nervous system itself. expression of normal cytokines in the brain is or but marked increases in several and cytokines after acute (see Hopkins & Rothwell, 1995). In experimental protein and bioactive IL-1, IL-6, TNF and IL-8 are increased in response to brain or experimental of inflammation such as lipopolysaccharide (see Hopkins & Rothwell, 1995). numerous clinical studies have shown increases in these and other cytokines in cerebrospinal fluid or brain at from with acute or neurological including and as well as infection and inflammation (see Hopkins & Rothwell, 1995). Increased expression of cytokines may of course be to brain and not directly in the However, studies on experimental animals indicate that some most notably IL-1 participate directly in acute (see Rothwell & 1993; Rothwell & Hopkins, 1995; Rothwell, 1996; Rothwell et al., 1996). The expression of IL-1β in response to injury or in rat brain within one hour of the almost in Increased protein for many days with expression also in and peripheral immune cells & Rothwell, unpublished data). to increase or brain IL-1 activity dramatically influence the which injection of than one of IL-1β by the from in the rat (Figure & Rothwell 1992; & Rothwell, 1996). This of is probably not due to the pyrogenic effects of IL-1, since it is not affected by cyclo-oxygenase inhibitors & Rothwell, unpublished data). Furthermore, intracerebroventricular injection of IL-6 also elicits fever, but in marked to IL-1, inhibits than & Rothwell, unpublished data). Effects of i.c.v. injection of recombinant vehicle IL-1ra or an on measured h after in the Means ± s.e.mean are shown, n = ***P < 0.001 vs of the or action of endogenous IL-1β by i.c.v. injection of an of which active or IL-1ra, reduces brain by (Figure & Rothwell 1996; et al., 1996). by IL-1ra is not associated with changes in body temperature or cardiovascular and, in is reduced in the as well as the & Rothwell, 1996). studies have that peripheral injection of doses of IL-1ra kg−1) also reduces by in the rat and even when after the IL-1ra also inhibits and increases the number of and neurological et al., 1995; et al., 1995 and see Rothwell et al., for the brain, the effects of IL-1 and IL-1ra on appear to be since IL-1 injected into the and but is when injected into the injection of IL-1ra the and from but does not when into the & Rothwell, 1997). These data indicate that IL-1 directly in brain but data show that inhibition of IL-1 also reduces brain by and in or injury fluid & Rothwell, or & Rothwell, et al., and clinical of experimental (see et al., 1996; Rothwell, 1996). IL-1 has been implicated in diverse forms of brain and may influence several which to (see Rothwell, 1996). The mechanisms of action of IL-1 are not known and may on the nature of the and probably effects of neurones, and brain cells (Figure seemingly effects of IL-1 have been for inhibition of entry and activity and of (see Hopkins & Rothwell, 1995; Rothwell, 1996). However, it that the effects of IL-1 are of interleukin-1 in the brain which may have or effects on IL-1 is not to in or in However, it or brain in through actions in the (see this like the hypothalamus, a of IL-1 as detected from the binding of biotinylated IL-1ra (Luheshi, Rothwell & Toulmond, unpublished data). IL-1 can also in and has been implicated directly in et al., 1996). Many effects of IL-1 in the brain are on and IL-1 is to with et al., 1996; & 1996). IL-1 can induce release of a of molecules from (eg and its protein and which may directly or other (see Rothwell, 1996; Rothwell et al., 1996). Furthermore, IL-1 could influence through effects on the it may influence blood brain barrier cause release of and induce expression of molecules to or (see Rothwell & 1993; et al., 1996; Rothwell, 1996; Rothwell et al., 1996). The mechanisms of IL-1 action on appear to be from those on fever. However, CRF has been implicated in both or brain causes of CRF mRNA at the site of injury and in the et al., 1995; & Rothwell, unpublished and administration of CRF receptor antagonists inhibits and brain injury in the rat (Figure et al., 1991; Strijbos et al., 1994; et al., 1995; & Rothwell, unpublished data). It is not yet known if the expression or actions of CRF in are related directly to IL-1, this is an hypothesis the relationship between these molecules in other responses to Effects of i.c.v. injection of a corticotrophin releasing factor (CRF) receptor antagonist on and brain in the rat. Data are from studies in which as of in their vehicle treated was measured h after days after fluid injury or h after infusion of an receptor CRF antagonist μg CRF or μg CRF, was injected i.c.v. immediately after the Means ± s.e.mean are shown, n = ***P < 0.001 vs vehicle of cytokine expression and action has of the involvement of these molecules in so many The more of their key in neuroimmune interactions has potential for modification of host defence responses to disease and for diverse neurological The most are inhibition of synthesis of proinflammatory cytokines or of their receptors. The may and the so little However, experimental data have several for neuroimmune including endogenous of cytokines cytokine binding receptors specific inhibitors of (eg or actions of cytokines and the of endogenous (eg on neuroimmune interactions and the of cytokines over the few may to novel for the of a number of peripheral and neurological disorders.
No takes yet. Share an insight, caveat, or question.
Nancy J. Rothwell (1997) studied this question.