1 Division of Immunaogy & Endocrinology National Institute for Biological Standards and Control (NIBSC), Blanche Lane,
South Mimms, Potters Bar, Hertfordshire, EN6 3QG, UK 2Brain Function Research Unit, School of Physiology, University of the Witwatersrand Medical School, York Road,
Parktown 2193, Johannesburg, South Africa
Abstract: Fever evoked by an exogenous pyrogen, either pathologically or by experimental interventions, is presumed to be mediated by endogenous pyrogenic cytokines, released from systemic mononuclear phagocytes when they interact with the pyrogen. The mechanism(s) by which peripherally elaborated cytokines transduce their pyrogenic message into central nervous system (CNS) signals still is a matter of vigorous debate. The current proposal is that proinflammatory cytokines, released into the circulation, communicate with the brain either directly or via the sensory circumventricular organs, to induce the synthesis and release of a more proximal mediator, prostaglandin E2, assumed to be the agent acting on thermoregulatory neurons. The pro-inflammatory cytokines act in sequence, with IL-6 the final member of the sequence. Several distinct pathways for cytokine signalling of the CNS have been reported, and, potentially, cytokines may engage a number of these pathways simultaneously, dependent upon the nature of the challenge (e.g. Gram-negative or Gram-positive pyrogen), the dose of the pyrogen and the compartment into which the pyrogen is presented (intravenously, intramuscularly, subcutaneously, intraperitoneally or intrathecally). The mechanisms that initiate, and those that sustain, fever likely differ, as various endogenous antipyretic systems, induced secondarily to the onset of fever, also determine the normal course of fever, although the effect of the combined influence of endogenous pyrogenic and antipyretic factors and the relative extent of each system's influence in different types of fever is still unknown.
From early on in the history of medicine, fever has been documented as a cardinal sign of disease, recognised, by physicians and patients alike, as an elevation of body temperature, or pyrexia. The thermal events in fever, however, constitute just one component, and not even an obligatory component, of a host response to insult, which also includes, for example, sickness behaviour, activation of the hypothalamic-pituitary-adrenal (HPA) axis and synthesis of acute-phase proteins (see for reviews, Kushner, 1988; Henderson et al„ 1998; Turnbull and Rivier, 1999). This suite of host responses, collectively termed the
»Corresponding author. Fax: +44 1707 646 730; E-mail: [email protected]
'acute-phase response' involves activation of numerous physiological, endocrinological and immunological systems.
The biochemical basis of fever has intrigued the human intellect since fever was recognised. Egyptian papyri dating back nearly 5000 years bear record of pus formation, a pathological process often accompanied by fever. It was only in the late 19th century, through the seminal treatise of William Welch (1888), that the programme for future investigations into the pathogenesis of fever was set. Welch associated fever with infection, and speculated that microbial agents produced fever through the release of 'ferments' (cytokines?), possibly from leukocytes, and that the 'ferments' acted directly on the brain, to initiate the peripheral changes responsible for the rise in body temperature, a hypothesis that was far ahead of its time. Ironically, co-incident with the emergence of this hypothesis, the discovery was made of bacterial endotoxin (Pfeiffer, 1892), a ubiquitous component of Gram-negative bacteria, subsequently found to be a potent inducer of cytokine synthesis. Since then, a number of exogenous pyrogens other than endotoxin has been shown to evoke fever, and numerous investigations have been carried out into the mechanisms of fever. Nevertheless, the investigations have continued to concentrate on endotoxin-induced experimental fever and, until recently, have been conducted, in the main, at the level of organ and organism (for reviews see, Atkins, 1960, 1984; Kluger, 1991; Dinarello and Bunn, 1997; Mitchell and Laburn, 1997; Zeisberger, 1999).
The brief account that follows engages the monumental body of research that has emanated in the last five or so years addressing the molecular basis of fever. The escalation of research has resulted mainly from the development of sophisticated molecular tools such as cloned cytokines and their receptors, receptor antagonists, soluble receptors, neutralising antibodies against cytokines and their receptor molecules, and knockout mice, in which the gene coding for a specific cytokine or its receptor has been deleted.
'It is tacitly assumed that fever is the product of a material fever-producing cause contained in the blood or tissue juice, the morbific action of which on the organism is antecedent to all functional disturbances whatsoever.'
John Bur don Sanderson, 1876
Research which started with the demonstration of an endogenous heat-labile factor present in the serum of rabbits during fever, the biological properties of which were quite separate from those of bacterial pyrogens (Beeson, 1948), and the pioneering work on endotoxin and fever orchestrated by Elisha Atkins (see for list of references, Atkins, 1984), led to the isolation and purification of endogenous pyrogen (Dinarello et al, 1974; Murphy et al, 1974; Cebula et al, 1979) and the discovery and subsequent cloning of the first cytokine, which, after many different guises, was named IL-1. Initially, it was assumed that IL-1 was the only endogenous pyrogen, and consequently the essential mediator of all fevers. However, several other cytokines such as tumour necrosis factor (TNF), IL-6, IL-8, leptin, interferons (IFNs) and leukaemia inhibitory factor (LIF), physically unrelated molecules but with biological effects remarkably similar to those of IL-1, also possess an intrinsic ability to evoke fever (Dinarello, 1996). The classic pro-inflammatory cytokines, IL-1, IL-6 and TNF, also appear to be the principal cytokines involved in fever genesis (Kluger, 1991).
The responses to experimental administration of exogenous pyrogens (e.g. Gram-negative and Grampositive bacteria) or their pyrogenic moieties often are indistinguishable from those following administration of endogenous pyrogens, such as IL-1 or TNF (Cannon et al, 1989), the febrile mediators released on exposure to exogenous pyrogen challenge. The physiological significance of the responses to doses of cytokines that far exceed endogenous concentrations is questionable, however (Kluger, 1991). For example, when a Gram-negative bacterium is administered into an experimentally constructed, subcutaneous airpouch, in guinea pigs, robust fever is evoked, and IL-6 concentrations increase in the preoptic area of the hypothalamus (POA), but circulating cytokine concentrations remain well below those necessary to induce fever when cytokines are administered exogenously (Ross et al, 2000). Differences in response to administration of exogenous pyrogens, or individual cytokines, at various doses, and via various routes, are well documented (Atkins, 1960; Kluger, 1991), and a clear relationship exists between the type, route and dose of pyrogen required to evoke fever, with the intravenous (i.v.) route, in rabbits at least, evoking fever most rapidly, followed by the intramuscular, subcutaneous and intraperitoneal (i.p.) routes, in that order (Cartmell et al, 2002). The potency of individual cytokines (gram for gram) in evoking fever differs also, with IL-6 being far less potent as an endogenous pyrogen than are IL-1 or TNF (see Helle et al, 1988; Dinarello et al, 1991; Dinarello, 1996), though synergism between the different cytokines (e.g. IL-1 and IL-6) does appear to occur (Dinarello, 1991; Stefferl et al, 1996; Cartmell et al, 2000). Nevertheless, in the genesis of fever, a distinct relationship between the pro-inflammatory cytokines appears to exist with TNF inducing IL-1 (Dinarello et al., 1986) and possibly IL-6, and IL-1 inducing IL-6 (see Billiau, 1988). In the periphery, these three cytokines are elevated in a regulated sequence in response to i.v. LPS administration, with TNF first, then IL-1 and finally IL-6 (Creasey et al., 1991; Givalois et al., 1994). Whether such an orderly pattern arises if a pyrogen first encounters the central nervous system (CNS) has yet to be determined. That any one of the pyrogenic cytokines alone is responsible for evoking fever in response to bacteria or their toxins, however, is unlikely in vivo (see Kluger, 1991). This redundancy is the result of the capacity of cytokines to influence the expression of other cytokines and their receptors, with an outcome that can potentiate or inhibit their actions, and may induce more distal co-mediators of cytokine-related bioactivities (e.g. prostaglandins, PGs) (Paul, 1989; Cohen and Cohen, 1996).
'They (the Gram-negative bacteria) display lipopolysaccharide endotoxin in their walls, and these macromolecules are read by our tissues as the very worst of bad news. When we sense lipopolysaccharide, we are likely to turn on every defense at our disposal.'
(Lewis Thomas, 1974) Cited in Henderson et al., 1998
Most experimental investigations of the molecular mechanisms of fever genesis have employed purified lipopolysaccharide (LPS), the glycolipid pyrogenic moiety of the Gram-negative bacterial membrane, to trigger the fever pathway. For this reason alone, we shall elaborate on cytokine synthesis and release during fever in the context of an LPS trigger. Two recent reviews (Cohen, 2002; Bochud and Calandra, 2003) on the pathogenesis of sepsis, a clinical condition resulting from a harmful host response to infection, are recommended to the interested reader for a clinical perspective on the fundamental principles governing bacterial-host interactions. Though they often erroneously are used interchangeably, the terms 'endotoxin' and 'LPS' are not synonymous: endotoxins are complexes of lipopolysaccharides, proteins, phospholipids and nucleic acids (Hitchcock et al., 1986). Peptidoglycan (PGN) and lipoteichoic acid (LTA) from Gram-positive bacteria also induce fever (see Zeisberger, 1999) and, like LPS, have the ability to activate nuclear transcription factor-icB (NFkB) signalling pathways and the production of cytokines (see for review, Nguyen et al., 2002). It has been argued that Gram-positive bacteria, though they routinely induce cytokine release from myeloid cells, might not require cytokine intermediates to evoke fever (see Mitchell and Laburn, 1997). The cytokines that are released, however, during Gram-positive fever presumably fulfil a role similar to their role in LPS-induced fever.
For LPS to exhibit full agonist potency in plasma, its lipid A domain (the pyrogenic moiety), it is believed, must bind to at least two non-signalling host accessory proteins: the constitutive serum protein LPS-binding protein (LBP), and soluble or membrane-bound CD 14. LBP, while present in plasma, normally is scarce in plasma-free peritoneal and other fluids (see Henderson et al., 1998), so LPS-CD14 interactions will not occur readily in the peritoneal cavity. Fever, however, is evoked in response to i.p. administration of LPS (see, Kluger, 1991; Zeisberger, 1999), implying an alternative signalling mechanism that activates macrophages in the peritoneal and perhaps other body compartments, and likely involves components of the complement cascade (see Sehic et al., 1998).
The LBP-LPS complex is no more active than is free LPS and the primary role of LBP is to function as a lipid-transfer protein, increasing the rate at which LPS interacts with soluble or membrane-bound CD14 (Pugin et al., 1993). Although CD14 does not have a domain for cytoplasmic signal transduction, its expression is required for optimal cell responses to LPS (Wright et al., 1990). Cellular responses to LPS are not solely dependent on CD 14, and several independent lines of evidence have supported the hypothesis that LPS interacts with a CD14-associated receptor to initiate the signalling process (see Ulevitch and Tobias, 1995). The literature on the LPS receptor itself is confusing, in part because of the heterogeneity of the LPS molecule, raising the prospect of non-specific cell-surface interactions, but mainly because, until recently, the signal-transducing receptor(s) for LPS had not been identified properly (see Henderson et al., 1998 for detailed review and list of references). There is consensus now that LPS initiates its pyrogenic
activities through a heteromeric receptor complex containing CD 14, together with the transmembrane protein Toll-like receptor (TLR) (Medzhitov et al, 1997; Poltorak et al, 1998), and at least one other protein, MD-2, which is essential to confer LPS responsiveness via its TLR (Shimazu et al, 1999). Figure 1 illustrates this concept. Lipopolysaccharide also can activate monocytes and macrophages via a CD14-independent pathway, a process apparently dependent on a plasma factor as yet unidentified (Cohen et al, 1995).
Toll-like receptors: the currency of pathogens?
Toll-like receptors are essential in the host defence against microbial pathogens. Ten distinct members
Fig. 1. Actions of IL-1 are mediated through IL-1RI that requires an accessory protein (IL-lRAcP) for signal transduction. The TLR-4 and TLR-2 signalling pathways, activated by Escherichia coli LPS and PGN, respectively, show remarkable similarities to the pathways employed by the IL-1 receptor. TLR2 also recognises lipoproteins and lipopeptides from several bacteria and does so via heterodimers formed between TLR2 and another TLR (TLR1 and TLR6). A secreted small molecule, MD-2, is essential for TLR-4 signalling. In all pathways, MyD88 is recruited to dimerised IL-1 R, TLR-4 or TLR-2 cytosolic domains and leads to the formation of the IRAK/TRAF-6 complex and subsequent activation of TAK-1, which serves as a branch point activating either IKK required for NFkB recruitment or upstream kinases that recruit p38 and JNK. Activation of the IkB kinase complex (IKK-aPy) also can be mediated by MAP kinase/ERK kinase 1 (MEKK1), bridged to TRAF-6 through another protein, termed "evolutionary conserved signalling intermediate in Toll pathways (ECSIT)', leading to phosphorylation of the inhibitory protein, IkBcx, allowing translocation of active NFkB into the nucleus, and subsequent activation of target genes. Cytokine induction through TLR4 and TLR2 depends also on the adapter molecule, TIRAP, whereas cytokine induction through IL-1R depends on MyD88, but not on TIRAP. TLR-4 signalling can occur also through a MyD88-independent pathway, regulated through the phosphorylation and nuclear translocation of IRF-3, and subsequent induction of IFN-p. Abbreviations: IKK-apy, IkB kinase complex; IL-1RI, IL-1 type I receptor; IRAK, IL-1 receptor-associated kinase; IRF-3, interferon regulatory factor 3; JNK, c-Jun N-terminal kinase; LBP, LPS-binding protein; LPS, lipopolysaccharide; Mai, MyD88-adapter-like; MEKK1, MAP kinase/ERK kinase 1; MyD88, myeloid differentiation factor 88; NFkB, nuclear factor-kb; PGN, peptidoglycan; TAK, TGFP-activated kinase; TIRAP, Toll/IL-1 receptor domain containing adapter protein; TLR, Toll-like receptor; TRAF6, TNF receptor-associated factor 6; Tollip, Toll-interacting protein.
of the mammalian TLR family (TLR 1-TLR 10) have been characterised thus far. All span the cell membrane, and appear to have been conserved throughout evolution: proteins homologous to Toll have been identified in a variety of organisms, including Drosophila and plants (reviewed by Akira, 2003). In addition, three other proteins, RP105, NODI and NOD2, share structural and functional homology with members of the Toll family (see for review, Vasselon and Detmers, 2002), and are proposed to act as intracellular receptors for invading bacteria and LPS (see Akira, 2003). Members of the TLR family are characterised by an extracellular domain containing leucine-rich repeats that differentially recognise 'pathogen-associated molecular patterns' shared by many pathogens (e.g. Gram-negative and Gram-positive bacteria), but not expressed by hosts (Janeway and Medzhitov, 2000). A large body of evidence indicates that TLR4 recognises LPS, whereas TLR2 recognises many different microbial components, including PGN from Staphylococcus aureus (see Akira, 2003), although the signal-transduction pathways that are activated are likely to differ (see for review, O'Neill et al, 2003). Toll-like receptor 4 appears to homodimerise, whereas for TLR2 ligands are recognised by a heterodimer of TLR2 and another TLR (TLR1, recognising bacterial lipopeptides or TLR6, recognising mycobacterial lipopeptides; refer to Fig. 1). These findings raise the possibility that a multiplicity of pathways originate from TLRs, with some convergent on NFkB and others with different targets. The functions of the other TLRs (TLR1, 3,5-10) are still under investigation and are not necessarily confined to activation by bacterial products (reviewed by Vasselon and Detmers, 2002; Akira, 2003). In vivo, it is likely that combinations of TLRs will be engaged by a pathogen and its products, leading to a refinement in the response seen in vitro (Underhill and Ozinsky, 2002).
families use similar signalling molecules (Fig. 1). Despite this similarity, individual TLRs recognise distinct structural components of pathogens and the signalling pathways evoked differ from one another and elicit different biological responses. As depicted in Fig. 1, activation of TLR by microbial components facilitates recruitment of IL-lR-associated kinase (IRAK) to TLR via the adaptor protein, myeloid differentiation factor 88 (MyD88). Recently, another adaptor molecule, TIR domain-containing adapter protein (TIRAP) or Mai, has been shown to have a crucial role in the MyD88-dependent signalling pathway shared by TLR2 and TLR4 signalling (O'Neill, 2002; Yamamoto et al., 2002), which is distinct from the role of MyD88 as a common adaptor (Fig. 1). Activated IRAK associates with TNF receptor-associated factor (TRAF) 6, and this association subsequently leads to the activation of two different pathways involving the Rel family transcription factor NFkB and the c-Jun N-terminal kinase (JNK)/p38 mitogen-activated protein (MAP) kinase family (Fig. 1). Nuclear translocation of NFkB and the activator protein 1 (AP-1) leads to transcriptional activation of numerous host-defence genes that encode cytokines, chemokines, proteins of the complement system, enzymes (such as cyclo-oxygenase (COX)-2 and the inducible form of nitric oxide (NO) synthase), adhesion molecules and immune receptors, and ultimately the local or systemic appearance of the host-defence molecules (see Fig. 1). Although TLR4 signalling can occur also through a MyD88-independent pathway (Kawai et al, 1999), responsible for the activation of IFN regulatory factor 3 (IRF-3) and the subsequent induction of IFN-P and IFN-inducible genes (Fig. 1), the MyD88-dependent pathway is essential for the inflammatory response mediated by LPS (see Akira, 2003).
Cytokine receptors and fever genesis
Inflammatory signalling pathways are initiated through the cytoplasmic Toll/IL-1 receptor (TIR) domain homologous to that in the IL-1 receptor (IL-1R) family (O'Neill, 2000). These receptor
Cytokines produce their selective biological effects by binding to specific membrane-bound receptors, thereby triggering a cascade of events, leading to either the MAP kinases/NFtcB or the Janus kinase-signal transducer and activator of transcription (JAK-STAT)-transduction pathways, and ultimately to specific patterns of gene activation (see for review and detailed list of references, Kishimoto et al, 1994; Henderson et al, 1998). More than 100 cytokines have now been identified and approximately 20,000 articles have been published on the topic of cytokine neurobiology alone. For the purposes of this chapter, the discussion will be restricted to the involvement in fever genesis of the three principal pyrogenic cytokines, that is TNF, IL-1 and IL-6.
Unlike the cytokines themselves, cytokine receptors share a number of structural similarities, allowing them to be grouped into superfamilies that use similar signal-transduction pathways. This similarity could in part explain the functional redundancy that occurs among cytokines, although IL-1, IL-6 and TNF, despite having many common biological activities, bind to distinct cell-surface receptors and do not share receptor subunits (see Kishimoto et al, 1994; Henderson et al, 1998). Most cytokine receptors consist of a multiunit complex, including a cytokine-specific ligand-binding component and a 'class'-specific signal-transduction unit (Sato and Miyajima, 1994).
The synthesis of TNF can be induced by a variety of stimuli including LPS, S. aureus, bacteria, viruses, fungi, protozoa, TNF itself, IL-1, IL-2, IFN, substance P, anti-T-cell reactivity antigen and tumour cells (see Mackowiak et al, 1997). Tumour necrosis factor, like IL-1, occurs in a and (3 forms. Tumour necrosis factor a is thought to be the main regulator of fever. Despite considerable overlap with the biological actions of IL-1, there is no apparent similarity in structure or in post-receptor events, between the IL-1 and TNF receptors (see, Dinarello, 1997). Tumour necrosis factor mediates its pleio-tropic effects by two structurally related, but functionally distinct, receptors: type I (TNFRI, p55) and type II (TNFRII, p75) (Bazzoni and Beutler, 1996). These two receptors differ in their transmembrane and cytoplasmic domains, congruent with them using separate signalling pathways, and different functions have been attributed to each of the receptors, although some redundancy has been described (see for review, Darnay and Aggarwal,
1997). Although both receptors are ubiquitously expressed in cells and interact with both forms of TNF, TNFRI is the most potent in inducing cytotoxic signals due to a 60 80-amino acid cytoplasmic sequence known as the death domain which is not present in TNFRII. The TNF receptor family lacks intrinsic signalling capacity and transduces signals by recruiting associating molecules such as the protein adaptor, TNFRI-associated death domain (TRADD) and the signalling molecules, TNF receptor-associated factor-2 (TRAF-2) and receptor-interacting protein (RIP) (Natoli et al, 1997; see for review, Guicciardi and Gores, 2003). The binding of TNF to its cognate p55 receptor (TNFRI) results in conformational changes in the receptor's intracellular domain leading to the rapid recruitment and formation of TRADD/TRAF-2/RIP complex and subsequent activation and translocation of NFkB (Fig. 2). The RIP/TRAF-2 complex involves also the MAPK cascade. Recruitment of Fas-associated protein with death domain (FADD) to the receptor promotes apoptosis (see Darnay and Aggarwal, 1997). In the case of TNFRII, signal transduction occurs via heterodimerisation of the receptor with TRAF1/TRAF2 and it is TRAF2 that activates NFkB signalling events. One of the most potent of all inducers of NFkB activity is the binding of TNF to its type I receptor (Baeuerle and Baltimore, 1996).
The synthesis of IL-1 can be induced by a variety of stimuli including LPS, IL-1 itself, TNF, IFNy and leukotrienes (see Mackowiak et al, 1997). The IL-1 family comprises two agonists, IL-1 a and IL-1 (5, and a highly selective, endogenous, IL-1 receptor antagonist (IL-lra). The three cytokines are secreted by similar cell types and in response to similar stimuli and IL-lra plays an important role in regulating endogenous IL-1 (see for review, Dinarello, 1996). Pharmacologically, IL-lra has been used extensively to investigate interactions between IL-1 and its receptor in a number of physiological systems (Dinarello and Thompson, 1991), fever included. In vivo, IL-1 a, IL-1 (3 and IL-lra are synthesised initially as precursors, of which pro-IL-la and
Was this article helpful?