General features

LPS are powerful stimuli of innate immune responses because they have no structural homologue in mammalian organisms (Beutler, Reference Beutler2000). They are released during bacteriolysis or during periods of rapid bacterial proliferation (Rietschel et al. Reference Rietschel, Schletter, Weidemann, El Samalouti, Mattern and Zahringer1998). LPS administration triggers a profound pro-inflammatory cytokine response (Abram et al. Reference Abram, Vu, Wraber and Doric2000) and, hence, mimics many features of the acute-phase response including the anorexia. LPS have been shown to induce taste aversions in different experimental situations (Langhans et al. Reference Langhans, Balkowski and Savoldelli1991; Weingarten et al. Reference Weingarten, Senn and Langhans1993), and conditioning may contribute to LPS anorexia under conditions that favour associative learning (Weingarten et al. Reference Weingarten, Senn and Langhans1993; Exton et al. Reference Exton, Bull and King1995). However, such learned effects do not appear to play a major role in the anorectic response to LPS under normal circumstances, i.e. when the illness is not associated with a novel diet (Weingarten et al. Reference Weingarten, Senn and Langhans1993). Fasting attenuates the feeding-inhibitory effect of LPS (Gautron et al. Reference Gautron, Mingam, Moranis, Combe and Laye2005), which may be related to the presumed lowering of the body-weight set point by inflammatory mechanisms (Lennie, Reference Lennie1998). While the exact mechanism(s) of this phenomenon are unknown, energy restriction apparently attenuates the stimulatory effect of LPS on macrophage cytokine production (Vega et al. Reference Vega, de Cabo and De Maio2004) and hypothalamic paraventricular nucleus activation (as evidenced by a reduction in LPS-induced c-fos expression; Gautron et al. Reference Gautron, Mingam, Moranis, Combe and Laye2005), both of which might curb anorexia in response to LPS. It is also interesting that LPS induces a stronger inhibition of feeding in females than in males (Geary et al. Reference Geary, Asarian, Sheahan and Langhans2004), a gender difference mainly related to oestrogen (Geary, Reference Geary2001; Geary et al. Reference Geary, Asarian, Sheahan and Langhans2004). While it is generally accepted that LPS induce anorexia by stimulating the production of cytokines that then act on the brain to inhibit feeding, there are several open questions, including: (1) where and in which cells the stimulation of cytokine production occurs; (2) which brain areas and neurochemicals mediate the resulting behavioural response.

Lipopolysaccharide receptor mechanisms

LPS trigger biological responses by LPS-binding protein-mediated binding to the cell-surface glycoprotein CD14 (Schutt et al. Reference Schutt, Bernheiden, Grunwald, Stelter, Menzel, Muller, Fan and Jack1999), which is present on many immune cells as well as on endothelial cells (EC). Furthermore, in cells devoid of CD14 the circulating soluble form of CD14 can replace membrane-bound CD14 (Akira, Reference Akira2000). LPS, CD14, the myeloid differentiation protein 2 and the toll-like receptor (TLR)-4, which is the ‘true’ LPS receptor (Akira, Reference Akira2000; Beutler, Reference Beutler2000), form the LPS receptor complex. TLR are a family of transmembrane proteins that mainly act as receptors for microbial substances (Akira, Reference Akira2000; Beutler, Reference Beutler2000). It has been shown that CD14 and TLR4, but not TLR2, are essential for the full expression of LPS anorexia (von Meyenburg et al. Reference von Meyenburg, Hrupka, Arsenijevic, Schwartz, Landmann and Langhans2004). LPS activation of TLR4 leads to recruitment of the myeloid adapter protein MyD88, which forms a complex with the Ser/Thr kinase IL-1 receptor-associated kinase that interacts with TNF receptor-activated factor 6. This process activates the transcription factors NF-κB and activating protein-1 (Beutler, Reference Beutler2002) and ultimately triggers the production of pro-inflammatory cytokines, prostanoids and other downstream mediators of LPS effects. The intracellular pathways of LPS and cytokine signalling overlap and have been extensively investigated. The absence of MyD88 signalling has recently been shown to completely eliminate anorexia in response to either LPS or IL-1β (Ogimoto et al. Reference Ogimoto, Harris and Wisse2006). Interference with NF-κB production has also been shown to block the feeding-inhibitory effect of IL-1β (Nadjar et al. Reference Nadjar, Bluthe, May, Dantzer and Parnet2005) and is the most likely mechanism by which genetic lack of PPARβ, administration of phosphodiesterase inhibitors such as pentoxifylline (Porter et al. Reference Porter, Hrupka, Altreuther, Arnold and Langhans2000) and some other pharmacological interventions antagonize the feeding-inhibitory effect of LPS (for review, see Langhans, Reference Langhans, Stricker and Woods2004). The neural mechanism(s) that mediate these effects, however, have not yet been determined. In summary, various experimental manipulations that interfere with TLR4 signalling and, hence, pro-inflammatory cytokine production, antagonize the feeding-inhibitory effect of LPS in animal models of systemic bacterial infections.

Role of cytokines

It is well known that cytokines orchestrate non-specific and specific immune reactions. They are broadly categorized as being pro-inflammatory or anti-inflammatory, i.e. they are involved in both the pathogenesis of signs of disease, such as anorexia and fever, and the host defence against the disease (for review, see Oppenheim, Reference Oppenheim2001). Several pro-inflammatory cytokines, such as IL-1, IL-2, IL-6, IL-8, IL-18, TNFα, interferon-γ (IFN-γ) and ciliary neurotrophic factor (CNTF), have been implicated in LPS anorexia. Each of these cytokines has been shown to inhibit eating after peripheral or central administration (for example, see Plata-Salaman, Reference Plata-Salaman1995; Langhans   Hrupka, Reference Langhans and Hrupka1999; Lambert et al. Reference Lambert, Anderson, Sleeman, Wong, Tan, Hijarunguru, Corcoran, Murray, Thabet, Yancopoulos and Wiegand2001; Netea et al. Reference Netea, Joosten, Lewis, Jensen, Voshol and Kullberg2006) and some of them are known to act synergistically (Yang et al. Reference Yang, Koseki, Meguid, Gleason and Debonis1994; Sonti et al. Reference Sonti, Ilyin and Plata-Salaman1996). The synergies are presumably related to the overlapping effects of the cytokines and to the fact that they act through converging intracellular signalling pathways. Leptin, which is not considered to be a classical cytokine, also affects immune functions and is in many aspects similar to cytokines (Sanchez-Margalet et al. Reference Sanchez-Margalet, Martin-Romero, Santos-Alvarez, Goberna, Najib and Gonzalez-Yanes2003). Leptin is also implicated in the feeding-inhibitory effect of LPS (see p. 323).

While genetic ablation of a particular cytokine or its receptor often does not substantially attenuate the anorectic effect of peripheral LPS (for review, see Langhans, Reference Langhans, Stricker and Woods2004), acute pharmacological or immunological antagonism of cytokines appears to be generally more effective (Bluthe et al. Reference Bluthe, Dantzer and Kelley1992; Porter et al. Reference Porter, Arnold and Langhans1998a, Reference Porter, Hrupka, Altreuther, Arnold and Langhans2000; Swiergiel   Dunn, Reference Swiergiel and Dunn1999; Laye et al. Reference Laye, Gheusi, Cremona, Combe, Kelley, Dantzer and Parnet2000; Harden et al. Reference Harden, du Plessis, Poole and Laburn2006). These seemingly discrepant findings may be related to the redundant and overlapping actions of the cytokines, which could permit unusually extensive developmental compensation. Accordingly, simultaneous interference with several cytokines often has a stronger effect on LPS anorexia than acute blockade of only one cytokine alone; indeed, such compound treatments are at times necessary in order to observe any effect (Swiergiel   Dunn, Reference Swiergiel and Dunn1999; Bluthe et al. Reference Bluthe, Laye, Michaud, Combe, Dantzer and Parnet2000), suggesting that pro-inflammatory cytokines can replace each other to a certain extent in mediating LPS anorexia.

Some data suggest a special role for IFN-γ in LPS anorexia (Arsenijevic et al. Reference Arsenijevic, Garcia, Vesin, Vesin, Arsenijevic, Seydoux, Girardier, Ryffel, Dulloo and Richard2000). IFN-γ is mainly produced in T-cells and natural killer cells (Billiau   Vendenbroeck, Reference Billiau, Vendenbroeck, Oppenheim and Feldmann2001), neither of which possesses TLR-4 (Beutler, Reference Beutler2002). Thus, LPS indirectly stimulates IFN-γ production through macrophage-derived IL-12 and IL-18 as well as TNFα (Doherty et al. Reference Doherty, Lange, Langstein, Alexander, Buresh and Norton1992; Billiau   Vendenbroeck, Reference Billiau, Vendenbroeck, Oppenheim and Feldmann2001). The main function of IFN-γ is to activate macrophages and EC, partly in synergy with macrophage-derived cytokines (Billiau   Vendenbroeck, Reference Billiau, Vendenbroeck, Oppenheim and Feldmann2001). Thus, by enhancing pro-inflammatory cytokine production and action, IFN-γ may be essential for the full expression of LPS anorexia (Arsenijevic et al. Reference Arsenijevic, Garcia, Vesin, Vesin, Arsenijevic, Seydoux, Girardier, Ryffel, Dulloo and Richard2000). In general, although it is clear that pro-inflammatory cytokines play a prominent role in mediating LPS anorexia, the complex interactions among several cytokines rather than the action(s) of any single cytokine appear to be crucial.

Roles of leptin and ghrelin

Leptin is another possible mediator of LPS anorexia. LPS and pro-inflammatory cytokines increase the expression and production of leptin in adipose tissue (Grunfeld et al. Reference Grunfeld, Zhao, Fuller, Pollock, Moser, Friedman and Feingold1996; Faggioni et al. Reference Faggioni, Fantuzzi, Fuller, Dinarello, Feingold and Grunfeld1998; Finck et al. Reference Finck, Kelley, Dantzer and Johnson1998), and there is a correlation between these increases and the feeding-inhibitory effects of the cytokines (Grunfeld et al. Reference Grunfeld, Zhao, Fuller, Pollock, Moser, Friedman and Feingold1996). In turn, leptin increases cytokine expression (Dixit et al. Reference Dixit, Schaffer, Pyle, Collins, Sakthivel, Palaniappan, Lillard and Taub2004). Moreover, neutralization of circulating leptin with a leptin antiserum has recently been shown to reverse the feeding-inhibitory effect of LPS (Sachot et al. Reference Sachot, Poole and Luheshi2004; Harden et al. Reference Harden, du Plessis, Poole and Laburn2006). Moreover, it was found that LPS causes an up-regulation of IL-1β and IL-1 receptor antagonist mRNA in the hypothalamus, and this effect is also attenuated by leptin antiserum (Sachot et al. Reference Sachot, Poole and Luheshi2004). These results suggest that leptin is a circulating mediator of LPS anorexia, possibly through a hypothalamic IL-1β-dependent mechanism. On the other hand, it has recently been reported (Gayle et al. Reference Gayle, Desai, Casillas, Beloosesky and Ross2006) that an anorectic dose of LPS given intraperitoneally barely increases plasma leptin in male rats despite a reduction in food intake. Studies in animals with genetic defects in the leptin system have also yielded mixed results. When compared with normal control animals LPS administration has a more pronounced feeding-suppressive effect in ob/ob (leptin-deficient) mice and it causes weaker food-intake inhibition in db/db (leptin receptor-deficient) mice (Faggioni et al. Reference Faggioni, Fuller, Moser, Feingold and Grunfeld1997) than in corresponding wild-type mice. Furthermore, a single intraperitoneal LPS injection reduces food intake similarly in lean (Fa/?) and obese (fa/fa) Zucker rats (Lugarini et al. Reference Lugarini, Hrupka, Schwartz, Plata-Salaman and Langhans2005). High doses of LPS (500 μg/kg or 1·0 mg/kg) also cause a similar initial (day 1) inhibition of feeding in lean and obese Zucker rats, although the recovery of normal food intake is somewhat delayed after the highest dose (1·0 mg/kg) in obese rats. In general, the available data suggest that leptin or functional leptin receptors are not necessary for the feeding-inhibitory effect of LPS, but that leptin nonetheless contributes to LPS anorexia in several ways.

The contribution of leptin to LPS anorexia might explain two seemingly-unrelated phenomena: (1) the attenuated feeding-inhibitory effect of LPS and pro-inflammatory cytokines after food deprivation; (2) the hypersensitivity of female individuals to LPS anorexia. As food deprivation reduces plasma leptin (Boden et al. Reference Boden, Chen, Mozzoli and Ryan1996), stimulation of leptin production by LPS in fasted individuals may fail to sufficiently increase circulating leptin for a feeding-inhibitory effect. As females appear to produce more leptin than males in response to LPS (Gayle et al. Reference Gayle, Desai, Casillas, Beloosesky and Ross2006) and are more sensitive to exogenous leptin than males (Clegg et al. Reference Clegg, Riedy, Smith, Benoit and Woods2003), leptin could well contribute to the stronger feeding-inhibitory effect of LPS in females compared with males.

Intraperitoneally administered LPS has been reported to substantially decrease circulating ghrelin (Basa et al. Reference Basa, Wang, Arteaga, Heber, Livingston and Tache2003; Hataya et al. Reference Hataya, Akamizu, Hosoda, Kanamoto, Moriyama, Kangawa, Takaya and Nakao2003; Wang et al. Reference Wang, Basa, Shaikh, Luckey, Heber, St Pierre and Tache2006). Interestingly, the LPS-induced decrease in plasma ghrelin is prevented by IL-1 receptor antagonist and indomethacin (Wang et al. Reference Wang, Basa, Shaikh, Luckey, Heber, St Pierre and Tache2006), suggesting that it is mediated by IL-1β and a prostanoid-dependent mechanism. Exogenous ghrelin antagonizes the LPS-induced inhibition of food intake and gastric emptying (Basa et al. Reference Basa, Wang, Arteaga, Heber, Livingston and Tache2003; Hataya et al. Reference Hataya, Akamizu, Hosoda, Kanamoto, Moriyama, Kangawa, Takaya and Nakao2003; Wang et al. Reference Wang, Basa, Shaikh, Luckey, Heber, St Pierre and Tache2006). Although it is unclear how LPS and cytokines inhibit gastric ghrelin production, it has been shown that ghrelin potently stimulates feeding (Tschop et al. Reference Tschop, Smiley and Heiman2000) and LPS inhibits feeding by reducing meal number (Langhans et al. Reference Langhans, Harlacher and Scharrer1989), which is at least consistent with an involvement of ghrelin.

Vagal afferents

Although IL-1β can activate vagal afferents (Niijima, Reference Niijima1996; Kurosawa et al. Reference Kurosawa, Uvnäs-Moberg, Miyasaka and Lundeberg1997), this mechanism does not appear to be crucial for the feeding-inhibitory effects of LPS and IL-1β. For example, subdiaphragmatic vagotomy has been reported to attenuate some cytokine-induced phenomena (Dantzer et al. Reference Dantzer, Konsman, Bluthe and Kelley2000), including inhibition of instrumental responses to obtain food induced by intraperitoneal LPS or IL-1β in mice (Bret-Dibat et al. Reference Bret-Dibat, Bluthe, Kent, Kelley and Dantzer1995). However, subdiaphragmatic vagal deafferentation, alone and in combination with celiac-superior mesenteric ganglionectomy, did not alter the anorexia after intraperitoneal injection of LPS or IL-1β in rats (Schwartz et al. Reference Schwartz, Plata-Salaman and Langhans1997; Porter et al. Reference Porter, Arnold and Langhans1998b). Subdiaphragmatic vagal deafferentation is the most selective and specific method available to lesion vagal afferents. These findings therefore show that abdominal vagal and spinal visceral afferents are not necessary for the anorectic effects of these immune stimuli, at least in the rat.

Blood–brain barrier mechanisms

Pro-inflammatory cytokines produced in response to LPS may directly act on the brain to elicit anorexia because they are actively transported across the blood–brain barrier (BBB) (Banks   Kastin, Reference Banks and Kastin1996) and also enter the brain where the BBB is ‘leaky’, i.e. in the circumventricular organs. Several lines of evidence suggest, however, that non-neural cells of the BBB are the most important site of cytokine action in response to LPS (see Fig. 1). BBB EC and perivascular cells such as microglia and macrophages, presumably together with blood monocytes, have emerged as a highly-complex regulatory interface controlling brain-mediated reactions to peripheral challenges (Licinio   Wong, Reference Licinio and Wong1997; Turrin   Rivest, Reference Turrin and Rivest2004). BBB EC and perivascular cells possess cytokine receptors (VanDam et al. Reference VanDam, DeVries, Kuiper, Zijlstra, DeBoer, Tilders and Berkenbosch1996; Deckert-Schluter et al. Reference Deckert-Schluter, Bluethmann, Kaefer, Rang and Schluter1999) as well as TLR (Laflamme   Rivest, Reference Laflamme and Rivest2001). In addition, membrane-bound CD14 is present on BBB EC under basal conditions (Lacroix et al. Reference Lacroix, Feinstein and Rivest1998), and a robust increase in CD14 mRNA levels takes place in these cells in response to a single peripheral injection of LPS (Lacroix et al. Reference Lacroix, Feinstein and Rivest1998; Rivest, Reference Rivest2003). Accordingly, peripheral administration of LPS or IL-1β leads to a rapid induction of c-fos mRNA in all non-neural cells of the BBB (Herkenham et al. Reference Herkenham, Lee and Baker1998). Furthermore, LPS and pro-inflammatory cytokines cause activation of the transcription factor NF-κB in BBB EC (Bierhaus et al. Reference Bierhaus, Chen, Liliensiek and Nawroth2000) and trigger the release of neuromodulators such as prostanoids or NO (Cao et al. Reference Cao, Matsumura, Yamagata and Watanabe1996; Nadeau   Rivest, Reference Nadeau and Rivest1999; Rivest, Reference Rivest1999). Activation of BBB EC and the feeding inhibition induced by IL-1β are both mediated by NF-κB activation. In a recent study (Nadjar et al. Reference Nadjar, Bluthe, May, Dantzer and Parnet2005) intracerebroventricular injection of a specific inhibitor of NF-κB activation was found to block the feeding-inhibitory effect of intraperitoneally-administered IL-1β and dramatically reduce IL-1β-induced c-fos expression in various brain regions. These findings strongly support the hypothesis that IL-1β-induced NF-κB activation at the BBB is a crucial step in the transmission of the immune signals mediating anorexia from the periphery to the brain.

Fig. 1. Schematic diagram of the interactions between monocytes and blood–brain barrier endothelial cells (BBB EC) in the transmission of the lipopolysaccharide (LPS)-induced immune signals causing anorexia. LPS acts on monocytes and BBB EC to stimulate the release of pro-inflammatory cytokines. In response to LPS and cytokines, BBB EC release monocyte chemoattractant protein-1 (MCP-1) which recruits additional monocytes. IL-12 and IL-18 trigger the release of interferon-γ (IFN-γ) from T-cells and natural killer (NK) cells. A major role of IFN-γ is to enhance the production and action of other pro-inflammatory cytokines. The combined action of LPS and cytokines on BBB EC finally activates cyclooxygenase-2 (COX-2) thus stimulating PGE₂ production. PGE₂ modulates neurons involved in control of food intake and energy balance. Perivascular cells (microglia, macrophages) and receptors (CD14, toll-like receptors, cytokine receptors) have been omitted for simplicity. For further details, see p. 323–324.

BBB EC also produce cytokines (Licinio   Wong, Reference Licinio and Wong1997; Bierhaus et al. Reference Bierhaus, Chen, Liliensiek and Nawroth2000) and show signal transducer and activator of transcription-3 activation in response to LPS (Rummel et al. Reference Rummel, Voss, Matsumura, Korte, Gerstberger, Roth and Hubschle2005). Furthermore, peripheral immune stimulation by LPS, IL-1β or TNFα triggers a rapid (within 30–90 min) increase in transcription of monocyte chemoattractant protein-1 in BBB EC and all circumventricular organs (Thibeault et al. Reference Thibeault, Laflamme and Rivest2001). Only ligands that trigger NF-κB signalling have the ability to increase monocyte chemoattractant protein-1 gene expression. The substantial increase in monocyte chemoattractant protein-1 production attracts monocytes that produce pro-inflammatory cytokines. The interaction of BBB EC and monocytes in response to LPS should thus lead to a substantial increase in local cytokine production and action (Fig. 1), which might explain several failures to relate cytokines in the systemic circulation and LPS anorexia.

An interesting feature of BBB EC is their polarization, with the luminal (blood-facing) and abluminal (brain-facing) cell membranes differing in their lipid, receptor and transporter compositions. Interestingly, constitutive and LPS-induced secretion of pro-inflammatory cytokines produced by BBB EC is polarized in favour of luminal secretion of these cytokines (Verma et al. Reference Verma, Nakaoke, Dohgu and Banks2006), whereas other mediators appear to be released abluminally (Fig. 1). Thus, despite the putative role of cytokines in LPS anorexia, it appears unlikely that BBB EC-derived cytokines act as messengers for neurons, suggesting that mediators downstream of pro-inflammatory cytokines, such as prostanoids and/or NO, fulfil this function.

Mediators downstream of cytokines

LPS and cytokines synergistically increase cyclooxygenase 2 mRNA expression in BBB EC and potently stimulate prostanoid, in particular PGE₂, production (DeVries et al. Reference DeVries, Hoogendoorn, Vandijk, Zijlstra, VanDam, Breimer, Vanberkel, DeBoer and Kuiper1995; Cao et al. Reference Cao, Matsumura, Yamagata and Watanabe1996; Rivest, Reference Rivest1999). Non-specific cyclooxygenase inhibitors attenuate the anorectic effects of LPS and IL-1β (Langhans et al. Reference Langhans, Harlacher and Scharrer1989, Reference Langhans, Savoldelli and Weingarten1993). Furthermore, the cyclooxygenase 2 inhibitor NS-398, but not the cyclooxygenase 1 inhibitor resveratrol, attenuated the anorectic effect of intraperitoneal LPS and blocked the concomitant LPS-induced increase in cerebrospinal PGE₂ (Lugarini et al. Reference Lugarini, Hrupka, Schwartz, Plata-Salaman and Langhans2002). Interestingly, Pecchi et al. (Reference Pecchi, Dallaporta, Thirion, Salvat, Berenbaum, Jean and Troadec2006) have recently shown a robust up-regulation of microsomal PGE synthase-1 enzyme in the brain in response to intraperitoneally- and intracerebroventricularly administered anorectic doses of IL-1β. Microsomal PGE synthase-1 catalyses the last step of PGE₂ biosynthesis, and its expression is stimulated by pro-inflammatory agents. In addition, IL-1β failed to decrease food intake in microsomal PGE synthase-1(-/-) mice (Pecchi et al. Reference Pecchi, Dallaporta, Thirion, Salvat, Berenbaum, Jean and Troadec2006), although these animals developed anorexia in response to an injection of PGE₂. Together these results demonstrate that microsomal PGE synthase-1 is essential for IL-1β-induced anorexia. All these findings are consistent with the notion that LPS and cytokines act on BBB EC (Fig. 1) to trigger the production and release of PGE₂, which acts on neurons that are involved in, or are connected to, brain sites involved in food-intake control.

Cytokines

Neurons in various brain areas increase expression of several pro-inflammatory cytokines, their accessory proteins and their receptors in response to peripheral administration of LPS (Gabellec et al. Reference Gabellec, Griffais, Fillion and Haour1995; Gayle et al. Reference Gayle, Ilyin and Plata-Salaman1997b; Turrin et al. Reference Turrin, Gayle, Ilyin, Flynn, Langhans, Schwartz and Plata-Salaman2001). While it appears unlikely that centrally produced cytokines are the exclusive mediators of peripheral LPS anorexia (for review, see Langhans, Reference Langhans, Stricker and Woods2004), they may contribute under certain circumstances. Leptin increases hypothalamic IL-1β, central injection of IL-1 receptor antagonist inhibited the hypophagic effect of central or peripheral injection of leptin and IL-1 receptor-knock-out mice did not reduce food intake in response to leptin (Luheshi et al. Reference Luheshi, Gardner, Rushforth, Loudon and Rothwell1999). These data suggest that hypothalamic IL-1β contributes to the feeding-inhibitory effect of leptin. As leptin is a possible mediator of LPS anorexia (see p. 323), central IL-1β might also be involved in peripheral LPS anorexia. More recently, Wisse et al. (Reference Wisse, Ogimoto and Schwartz2006) have shown that the melanocortin antagonist SHU9119 blunts the LPS-mediated increase in hypothalamic IL-1β, but that pharmacological or genetic disruption of IL-1 receptor signalling does not prevent the anorexia induced by the melanocortin agonist MTII. These data question the role of central IL-1β as a major mediator of LPS anorexia because SHU9119 (Huang et al. Reference Huang, Hruby and Tatro1999) and genetic lack of the melanocortin-4 receptor (Marks et al. Reference Marks, Ling and Cone2001) attenuated the feeding-suppressive effect of LPS. Other direct tests of the role of central IL-1β have also yielded inconsistent results; while intracerebroventricular administration of IL-1 receptor antagonist failed to inhibit the feeding-inhibitory effect of intraperitoneal LPS in rats (Bluthe et al. Reference Bluthe, Dantzer and Kelley1992), it did attenuate the effect of intraperitoneal LPS in mice (Laye et al. Reference Laye, Gheusi, Cremona, Combe, Kelley, Dantzer and Parnet2000). The reason for this discrepancy is unclear.

CNTF, a trophic factor for motor neurons in the ciliary ganglion and spinal cord, has been found to markedly reduce food intake and body weight (see Lambert et al. Reference Lambert, Anderson, Sleeman, Wong, Tan, Hijarunguru, Corcoran, Murray, Thabet, Yancopoulos and Wiegand2001). IL-1β is essential for CNTF production in response to brain injury or trauma (Herx et al. Reference Herx, Rivest and Yong2000), raising the possibility that CNTF may also be a downstream mediator of IL-1β effects on food intake. Some data suggest that CNTF ultimately affects energy balance by reducing the expression and action of neuropeptide Y (Xu et al. Reference Xu, Dube, Kalra, Farmerie, Kaibara, Moldawer, Martin and Kalra1998). A reduction in hypothalamic neuropeptide Y mRNA has also been reported during IL-1β-induced anorexia (Gayle et al. Reference Gayle, Ilyin and Plata-Salaman1997a), although the decrease in neuropeptide Y expression appears to be too small to account for the substantial reduction in food intake. It is possible, however, that cytokine-induced decreases in neuropeptide Y attenuate the feeding that normally occurs in response to an energy deficit (Inui, Reference Inui1999). Recently, it has been reported (Steinberg et al. Reference Steinberg, Watt, Fam, Proietto, Andrikopoulos, Allen, Febbraio and Kemp2006) that a potent CNTF analogue and leptin reduce food intake and hypothalamic AMP kinase expression similarly. Numerous reports in the last few years suggest that hypothalamic AMP kinase functions as an energy sensor in the control of energy balance (Small et al. Reference Small, Carling and Bloom2004).

Serotonin

The increase in c-fos expression (i.e. the activation) in medullary and hypothalamic paraventricular neurons that is elicited by LPS or cytokine treatment (Elmquist   Saper, Reference Elmquist and Saper1996; Ericsson et al. Reference Ericsson, Arias and Sawchenko1997; Lacroix   Rivest, Reference Lacroix and Rivest1997) appears to be mediated by PGE₂ (Ericsson et al. Reference Ericsson, Arias and Sawchenko1997; Lacroix   Rivest, Reference Lacroix and Rivest1997). This finding is interesting because serotonin (5-HT) and catecholamine cell groups in the midbrain and hindbrain, but not in the paraventricular nucleus, possess PG EP₃ receptors and are activated by PGE₂ (Ericsson et al. Reference Ericsson, Arias and Sawchenko1997; Nakamura et al. Reference Nakamura, Li, Kaneko, Katoh and Negishi2001). Serotonergic projections from the midbrain raphe area and the hindbrain to the hypothalamus are particularly interesting candidate pathways for the anorectic effects of LPS and cytokines. First, it has recently been observed (BS Kopf, N Geary, W Langhans and L Asarian, unpublished results) that intraperitoneal LPS increases c-fos in large parts of the raphe. Second, it has also been found (see Langhans, Reference Langhans, Stricker and Woods2004) that microinjection of a cyclooxygenase 2 inhibitor into the dorsal raphe nucleus markedly reduces anorexia following intraperitoneal LPS, and that microinjection of PGE₂ into the same area decreases food intake. The dorsal raphe nucleus also contains IL-1 receptors (Cunningham et al. Reference Cunningham, Wada, Carter, Tracey, Battey and Desouza1992), and central as well as peripheral administration of IL-1β and TNFα increased serotonergic activity in this area (Clement et al. Reference Clement, Buschmann, Rex, Grote, Opper, Gemsa and Wesemann1997). 5-HT potently inhibits eating, apparently mainly through the 5-HT_1B and/or 5-HT_2C receptors (Simansky, Reference Simansky1995; Nonogaki et al. Reference Nonogaki, Strack, Dallman and Tecott1998). Pretreatment with a highly-specific 5-HT_2C receptor antagonist blocked the anorexia induced by both peripheral and central injection of LPS or IL-1β in rats (von Meyenburg et al. Reference von Meyenburg, Langhans and Hrupka2003a,b). Furthermore, administration of the 5-HT_1A autoreceptor agonist 8-hydroxy-2-di-n-propylamino-tetralin directly into the dorsal raphe nucleus blocked the feeding-inhibitory effect of peripheral LPS and IL-1β (von Meyenburg et al. Reference von Meyenburg, Langhans and Hrupka2003a,b), whereas pharmacological antagonism of other 5-HT receptors (5-HT_1B, 5-HT_2A, 5-HT₃) did not. Pharmacological 5HT_2C antagonism also attenuated the feeding-inhibitory effect of intraperitoneal LPS in mice, although LPS did not reduce food intake in 5-HT_2C-knock-out mice (Asarian et al. Reference Asarian, Kopf, Geary and Langhans2007). In addition, some pharmacological data in mice (Swiergiel   Dunn, Reference Swiergiel and Dunn2000) suggest that the involvement of central 5-HT neurons in LPS and cytokine-induced anorexia is situationally variable. In summary, therefore, although several findings implicate the 5-HT neurons in the median raphe nucleus and 5-HT_2C receptors in the hypothalamus in LPS anorexia, it is not yet clear whether this mechanism plays a necessary role, at least in mice. Finally, emerging evidence indicates that 5-HT modulates the release of endogenous agonists and antagonists of brain melanocortin receptors, which are crucial for the central control of energy balance (Heisler et al. Reference Heisler, Cowley, Tecott, Fan, Low and Smart2002, Reference Heisler, Jobst, Sutton, Zhou, Borok and Thornton-Jones2006).

Neuropeptides

LPS increases the number of glucagon-like peptide-1 neurons in the nucleus of the solitary tract that express c-fos (Rinaman Reference Rinaman1999), and both 3rd-intracerebroventricular and 4th-intracerebroventricular administration of the glucagon-like peptide-1 receptor antagonist exendin-(9–39) attenuated the anorectic response to intraperitoneal LPS in rats (Comer   Rinaman Reference Comer and Rinaman2000; Grill et al. Reference Grill, Carmody, Sadacca, Williams and Kaplan2004). These findings have recently been extended by Grill et al. (Reference Grill, Carmody, Sadacca, Williams and Kaplan2004) to show that 3rd-intracerebroventricular administration of exendin is ineffective when the caudal flow of cerebrospinal fluid is blocked by occlusion of the cerebral aqueduct, which suggests that LPS anorexia is mediated in part by release of glucagon-like peptide-1 within the caudal brain stem.

Recent findings (Becskei et al. Reference Becskei, Hernadfalvy, Arsenijevic, Lutz and Langhans2006) suggest that peripheral LPS reduces c-fos expression in the lateral hypothalamic area and decreases the number of lateral hypothalamic area neurons expressing orexin-A protein in mice deprived of food for 12 h. As orexin-A has a potent orexigenic effect (Rodgers et al. Reference Rodgers, Ishii, Halford and Blundell2002), this finding raises the possibility that a decrease in orexin-A expression contributes to LPS anorexia. Since orexin-A is mainly implicated in arousal (Rodgers et al. Reference Rodgers, Ishii, Halford and Blundell2002), the inhibitory effect of LPS on orexin-A protein-expressing neurons might also be involved in LPS-induced lethargy and inactivity.

Peripheral injection of IL-1β increased hypothalamic corticotrophin-releasing factor mRNA (Suda et al. Reference Suda, Tozawa, Ushiyama, Sumitomo, Yamada and Demura1990), and IL-1β-induced anorexia was attenuated by 3rd-intracerebroventricular administration of a corticotrophin-releasing factor antagonist (Uehara et al. Reference Uehara, Sekiya, Takasugi, Namiki and Arimura1989), suggesting that hypothalamic corticotrophin-releasing factor is involved in IL-1β-induced anorexia. Prostanoids mediate the effect of IL-1β on hypothalamic corticotrophin-releasing factor release (Watanabe et al. Reference Watanabe, Morimoto, Sakata and Murakami1990), suggesting a link between the putative role of PGE₂ (Langhans et al. Reference Langhans, Savoldelli and Weingarten1993; Lugarini et al. Reference Lugarini, Hrupka, Schwartz, Plata-Salaman and Langhans2002) and corticotrophin-releasing factor (Uehara et al. Reference Uehara, Sekiya, Takasugi, Namiki and Arimura1989) in the anorectic effects of LPS and IL-1β.

Finally, LPS stimulates the release of α-melanocyte-stimulating hormone (Catania et al. Reference Catania, Suffredini and Lipton1995), which antagonizes acute-phase reactions at various levels (cytokine production, cytokine action; Lipton   Catania, Reference Lipton and Catania1998). α-Melanocyte-stimulating hormone binds to central melanocortin receptors (MCn-R; MC3-R and MC4-R), and central administration of MC4-R agonists inhibits food intake, increases energy expenditure and reduces body weight (see Tritos   Maratos-Flier, Reference Tritos and Maratos-Flier1999). In turn, deficiency of the MC4-R is associated with increases in food intake and body weight (Fan et al. Reference Fan, Boston, Kesterson, Hruby and Cone1997). Intracerebroventricular administration of α-melanocyte-stimulating hormone enhanced, and intracerebroventricular administration of the MC3-R and MC4-R antagonist SHU9119 attenuated, intraperitoneal LPS anorexia in rats (Fan et al. Reference Fan, Boston, Kesterson, Hruby and Cone1997; Huang et al. Reference Huang, Hruby and Tatro1999). More recently, the anorexia induced by intraperitoneal LPS and cytokines has also been shown to be attenuated by the endogenous MC3-R and MC4-R antagonist Agouti-related peptide and in MC4-R-knock-out mice but not in MC3-R-knock-out mice (Marks et al. Reference Marks, Butler, Turner, Brookhart and Cone2003). These results specifically implicate the central MC4-R in LPS and cytokine-induced anorexia. Given the putative role of 5-HT in the anorectic effects of LPS and IL-1β (see p. 325), it is interesting to note that the melanocortin system is a downstream target of serotonergic neurons (Heisler et al. Reference Heisler, Cowley, Tecott, Fan, Low and Smart2002, Reference Heisler, Jobst, Sutton, Zhou, Borok and Thornton-Jones2006).