Central mediators involved in the febrile response: effects of antipyretic drugs

Abstract

Fever is a complex signal of inflammatory and infectious diseases. It is generally initiated when peripherally produced endogenous pyrogens reach areas that surround the hypothalamus. These peripheral endogenous pyrogens are cytokines that are produced by leukocytes and other cells, the most known of which are interleukin-1β, tumor necrosis factor-α, and interleukin-6. Because of the capacity of these molecules to induce their own synthesis and the synthesis of other cytokines, they can also be synthesized in the central nervous system. However, these pyrogens are not the final mediators of the febrile response. These cytokines can induce the synthesis of cyclooxygenase-2, which produces prostaglandins. These prostanoids alter hypothalamic temperature control, leading to an increase in heat production, the conservation of heat, and ultimately fever. The effect of antipyretics is based on blocking prostaglandin synthesis. In this review, we discuss recent data on the importance of prostaglandins in the febrile response, and we show that some endogenous mediators can still induce the febrile response even when known antipyretics reduce the levels of prostaglandins in the central nervous system. These studies suggest that centrally produced mediators other than prostaglandins participate in the genesis of fever. Among the most studied central mediators of fever are corticotropin-releasing factor, endothelins, chemokines, endogenous opioids, and substance P, which are discussed herein. Additionally, recent evidence suggests that these different pathways of fever induction may be activated during different pathological conditions.

Abbreviations

AH/POA

anterior hypothalamus/preoptic area

BAT

brown adipose tissue

CCL

CC chemokine ligand

CCR

CC chemokine receptor

CINC-1

cytokine-induced neutrophil chemotactic factor-1

COX-2

cyclooxygenase-2

CRF

corticotropin-releasing factor

CXCL

CXC chemokine ligand

CXCR

CXC chemokine receptor

EOp

endogenous opioid

ET

endothelin

HLI

heat loss index

IFN

interferon

IL

interleukin

LPS

lipopolysaccharide

4-MAA

4-methylaminoantipyrine

MCP

monocyte chemotactic protein

MIP

macrophage-inflammatory protein

mPGES1

microsomal prostaglandin E synthase-1

NK

neurokinin

NSAID

nonsteroidal antiinflammatory drug

OVLT

organun vasculosum laminae terminalis

nor-BNI

nor-binaltorphimine

PAF

platelet-activating factor

PFPF

preformed pyrogenic factor

PG

prostaglandin

Poly I:C

polyinosinic-polycytidylic acid

RANKL

receptor activator of nuclear factor κB ligand

RANTES

regulated on activation, normal T cells expressed and secreted

SAL

saline

SEM

standard error of the mean

Tb

body temperature

TLR

Toll-like receptor

TNF-α

tumor necrosis factor-α

Tsv

Tityus serrulatus venom

Ucn

urocortin

VMH

ventromedial hypothalamus

Introduction

Endothermic animals have complex mechanisms to maintain body temperature between narrow limits. However, under some conditions (e.g., during infectious diseases), adjustments in body temperature are necessary. The febrile response is a controlled increase in body temperature that is caused by elevation of the hypothalamic set point by endogenous mediators that are produced during such conditions.^1,2 Specifically, the preoptic area in the anterior hypothalamus (AH/POA) controls thermoregulatory mechanisms that are responsible for normal temperature maintenance and changes that occur during the febrile response.^3,4

The administration of lipopolysaccharide (LPS) from Gram-negative bacteria in laboratory animals is the classic model of fever induction. Therefore, most knowledge of fever-inducing mechanisms is associated with this pathogen-associated molecular pattern. Lipopolysaccharide stimulates Toll-like receptors (TLRs), specifically TLR4, in many cell types, including macrophages,⁵ adipocytes,^6,7 Kupffer cells,^8,9 and microglia,^10,11 inducing the release of cytokines that deliver messages to the AH/POA to induce fever.¹ However, other molecules, such as zymosan, imiquimod, ODN1668, and polyinosinic-polycytidylic acid (Poly I:C), can also stimulate other TLRs. Among the released cytokines that can induce fever, usually called endogenous pyrogens, are interleukin-1β (IL-1β), IL-1α, tumor necrosis factor-α (TNF-α), IL-6, CXCL8/IL-8, CXCL1/cytokine-induced neutrophil chemotactic factor-1 (CINC-1), macrophage-inflammatory protein-1α (MIP-1α), MIP-1β, interferon β (IFN-β), IFN-γ, and receptor activator of nuclear factor κB ligand (RANKL).^1,3,12-15 Some of these cytokines, including CXCL8/IL-8, CXCL1/CINC-1, CCL3/MIP-1α, and RANKL, appear to be generated in the central nervous system.^12-15 and are discussed herein, whereas others are generated in the periphery. In addition to cytokines, evidence also indicates that platelet-activating factor (PAF), a lipid mediator that is produced during inflammation, can induce fever. Steiner and Romanovsky showed that peripherally infused PAF was able to induce fever in rats.¹⁶ Some of these mediators clearly do not participate in the LPS-induced febrile response, such as CCL3.¹⁷ and PAF.¹⁶ Once these mediators are generated in peripheral sites, several hypotheses have been proposed to explain how they reach the AH/POA to cause fever.^18-23 However, the participation of TLRs that are expressed in the central nervous system should also be acknowledged. Compelling evidence suggests the presence of some of these receptors (e.g.,, TLR2, TLR3, TLR4, TLR5, and TLR7) in the central nervous system.^24,25 Nakano et al. recently reported an increase in the expression of TLR4 in astrocytes and microglia in circumventricular organs.²⁶ Additionally, using chimeric mice that expressed TLR4 in the central nervous system resident cells versus in haematopoietic cells, the expression of this TLR in the central nervous system resident cells was shown to be necessary for sustaining the expression of inflammatory mediators in the brain.²⁷ Several studies have also shown that microcultures of OVLT cells released TNF-α and IL-6 in response to LPS and Poly I:C.^10,11,28 Furthermore, cells from these circumventricular organs respond to these cytokines. Wuchert et al. showed that IL-1β and TNF-α increased calcium transients in cells (i.e., astrocytes and neurons) from the area postrema.²⁹ Neurons from the subfornical organ also responded to IL-1β, with depolarization and an increase in spike frequency.³⁰ Altogether, these studies suggest that these TLR agonists can stimulate the synthesis of cytokines in the central nervous system, particularly in circumventricular organs, and the cytokines that are generated can stimulate these cells. Peripherally or centrally generated cytokines can induce other central mediators near the AH/POA, leading to changes in the firing of neurons that effectively activate heat production and conservation mechanisms, in addition to inhibiting heat loss.

The aim of the present review is to discuss the most known of these mediators that are generated in the central nervous system. We do not focus on cytokines that can be generated in both the periphery and central nervous system. We first discuss the importance of prostaglandins (PGs) in the febrile response that is induced by different exogenous and endogenous pyrogens and how different known antipyretic drugs affect the increase in this mediator in the hypothalamus. We also reexamine the current knowledge of the involvement of other central mediators of fever, such as corticotropin-releasing factor (CRF), endothelins (ETs), chemokines, endogenous opioids (EOps), and substance P.

Prostaglandins

In 1971, Milton and Wendlandt proposed that the antipyretic activity of nonsteroidal antiinflammatory drugs (NSAIDs) was related to their ability to inhibit the synthesis and release of PGs that are induced by endogenous pyrogens, in which the injection of several E-series PGs induced fever in cats and rabbits.³¹ They also reinforced these conclusions by showing that PGE₁, when injected in the third ventricle in cats, induced a dose-related febrile response that was unaffected by acetaminophen. In the same year, Feldberg and Saxena showed that PGE₁, when injected intracerebroventricularly, also induced fever in rats and rabbits, and high levels of PG were found in cerebrospinal fluid that was collected from the third ventricle in febrile cats.³² Later, radioimmunoassays were performed during the febrile response, revealing an increase in PGE₂ levels in the brains of several species.^33-35 Moreover, peripheral IL-1β administration in rats increased PGE₂ levels in several hypothalamic areas, particularly in the organun vasculosum laminae terminalis (OVLT), one of the circumventricular organs that lacks a blood-brain barrier and is localized near the AH/POA.^36,37

Another arachidonic acid metabolite that is involved in the genesis of the febrile response is PGF_2α. This prostanoid also increases body temperature in rats and rabbits when administered intracerebroventricularly.^38-40 An increase in the concentration of this PG was also observed in cerebrospinal fluid in rats that received peripheral endotoxin administration.⁴¹

However, the precise site of the production of PGs during the febrile response has been a matter of debate for many years. Katsuura et al.⁴² showed that the endogenous pyrogen IL-1β increased PGE₂ levels in cultures of astrocyte cells from rats, suggesting that glial cells could be the source of these eicosanoids during fever. Later, immunoreactive cyclooxygenase-2 (COX-2), the key enzyme that is involved in the production of these eicosanoids, was identified in dendrites and cell bodies of neurons in several brain areas, particularly the paraventricular nucleus of the hypothalamus and AH/POA in naive animals.⁴³ Recently, endothelial cells in the brain were identified as PGE₂-producing cells that are critical for the progression of LPS-induced fever. Mice that had selective deletion of COX-2 and microsomal prostaglandin E synthase 1 (mPGES1; a sequential enzyme for PGE₂ production after COX-2) in brain endothelial cells.⁴⁴ and mice that expressed mPGES-1 only in haematopoietic cells did not develop fever in response to this pyrogenic stimulus.⁴⁵

Since 1990, the OVLT and AH/POA have been considered the hypothalamic sites that are responsible for the production and action of PGs, respectively, during fever.^36,37 However, another area that is sensitive to the action of these eicosanoids is the ventromedial hypothalamus (VMH).⁴⁶ The VMH is an important site for the control of the autonomic nervous system and particularly brown adipose tissue (BAT) thermogenesis.⁴⁷ Later, several authors confirmed that the effects of PGs on thermogenesis and increase in body temperature depend on the sympathetic activation of heat production in BAT, likely through the central stimulation of sympathetic flow through the solitary tract and VMH.^48,49 The concentration of PGE₂ in the POA/AH may also be subjected to endogenous antiinflammatory control mechanisms. For example, the febrile response that was induced by intravenous LPS administration in rabbits was reduced by the antiinflammatory cytokine IL-10, although no changes in TNF-α levels were observed, suggesting that IL-10 reduces fever by controlling the production of PGE₂.⁵⁰

However, COX-2 mRNA expression was not found in the POA/HA but rather in blood vessels and leptomeninges that surrounded this area, suggesting that this would be the site that is responsible for PG production during the febrile response.^51,52 Later, Cao et al.^18,53 showed that systemic IL-1β and TNF-α injections also induced COX-2 mRNA expression in blood vessels in rats.

Eskilsson et al. showed that mPGES-1 is expressed in the mouse brain during non-immune challenge conditions.⁴⁵ Following immune stimulation, mPGES-1 expression increased only in brain endothelial cells that co-expressed COX-2.⁴⁵ These results are in agreement with Gaetano et al.,⁵⁴ who reported increases in COX-2 and mPGES-1 expression in the POA/AH following intravenous LPS injections in mice, whereas no other enzymes that are related to PG synthesis were altered. COX-2 and mPGES-1 knockout mice did not develop a febrile response after LPS injection.^44,55,56 In rats, the basal expression of mPGES-1 in the brain is lower than in mice, but its expression increased after LPS or IL-1β injection, and mPGES-1 was co-expressed with COX-2.^57,58 Similarly, Rummel et al.⁵⁹ found that LPS injection into the air pouch increased mPGES-1 mRNA expression in the rat brain, and this induction was suppressed by anti-IL-6 antiserum.

Engstron et al. generated chimeric mice that expressed mPGES-1 in either haematopoietic cells or non-haematopoietic cells.⁶⁰ Only animals with mPGES-1 expression in non-haematopoietic cells developed fever or an increase in PG concentration in cerebrospinal fluid after LPS administration. Animals with mPGES-1 expression only in haematopoietic cells exhibited elevated PGE₂ levels in plasma, but they did not develop fever, suggesting that during fever PGE₂ is formed in the brain.⁶⁰ However, because the febrile response was not completely abolished in this study, the authors suggested that additional sources can synthesize PGs. In fact, Steiner et al. showed that peripherally derived PGE₂ is also involved in LPS-induced fever. They constructed mouse chimeras and found that all 3 phases of the typical LPS-induced febrile response depended on TLR4 signaling. The first phase was triggered by TLR4 in haematopoietic cells. The second and third phases depended on TLR4 signaling in both haematopoietic and non-haematopoietic cells.⁶¹ These authors subsequently showed that PGE₂ synthesis (phosphorylation of cytosolic phospholipase A₂ and upregulation of COX-2) occurred in macrophages in the liver and lungs during the onset of fever.⁶² Moreover, the immune neutralization of plasma PGE₂ delayed the initiation of fever.⁶² Consistent with these studies, Kupffer cell depletion prevented the increase in peripheral PGE₂ and rise in body temperature that were induced by LPS in guinea pigs, suggesting that the peripheral generation of this eicosanoid by these cells is also important for fever.⁶³

Recently, Eskilsson et al. generated mice with selective deletion of IL-6 receptor α in several cell types, and only the deletion of this receptor in brain endothelial cells attenuated the LPS-induced febrile response.⁶⁴ Importantly, this deletion also attenuated the induction of COX-2 in the hypothalamus, suggesting that these cells are responsible for the generation of PGs during fever.⁶⁴ Corroborating these findings, brain endothelial cells are PGE₂-producing cells that are critical for the progression of LPS-induced fever. Mice with selective deletion of COX-2 and mPGES1 in brain endothelial cells.⁴⁴ and mice that expressed mPGES-1 only in haematopoietic cells did not develop fever in response to this pyrogenic stimulus.⁴⁵

PGE₂ may activate 4 types of G-protein-coupled receptors: EP₁, EP₂, EP₃, and EP₄. However, only EP₁, EP₃, and EP₄ have been identified in the AH/POA.^65,66 In this region, EP₃ receptors are found in many neuron cell bodies and dendrites. Although EP₁ and EP₄ mRNA expression has also been detected in the AH/POA,⁶⁵ only EP₃ receptor knockout mice did not develop fever in response to PGE₂, IL-1β, or LPS.^67,68 Moreover, EP₃ receptor knockout in neurons in the median POA and medial POA also reduced the febrile response that was induced by PGE₂ and LPS.⁶⁶ The neural circuitry that is activated after EP₃ receptor activation to induce fever has been reviewed previously.^69,70

Several studies have also shown that selective COX-2 inhibitors act as antipyretics in rats.^71-73 A selective COX-2 inhibitor also effectively reduced the LPS-induced febrile response in nonhuman primates, confirming the involvement of this enzyme and PGs in the progression of fever in this species.⁷⁴

These eicosanoids have also been shown to be involved in the febrile response that is induced by other exogenous stimuli, such as Poly I:C,⁷² Staphylococcus aureus,⁷³ and complex infectious conditions such as sepsis induced by cecal ligation and puncture.⁷⁵ The participation of these eicosanoids has also been suggested after stimulation with the TLR7 agonist imiquimod and TLR9 agonist ODN1668, which increased the expression of enzymes that are involved in PGE₂ synthesis in the rat hypothalamus.^76,77

With regard to endogenous pyrogens, the febrile response that is induced by IL-1β, TNF-α, IFN-γ, IL-6,^2,3 ciliary neurotrophic factor,⁷⁸ IL-11,⁷⁹ CCL5/RANTES (regulated on activation, normal T cells expressed and secreted),⁸⁰ CXCL1/CINC-1,¹⁴ and CC chemokine ligand 22.⁸¹ depends on PG synthesis, in which the febrile response that is induced by these pyrogens is blocked by COX inhibitors. Conversely, the febrile response that is induced by some chemokines, such as CXCL8/IL-8.¹² in rats and CCL3/MIP-1α in rabbits and rats, is not reduced by treatment with COX inhibitors, suggesting that these chemokines induce fever through mechanisms that are independent of PG synthesis.^82,83 The role of these and other chemokines as central mediators of fever are discussed in more detail below.

We have also shown that low doses of the selective COX-2 inhibitor celecoxib or nonselective COX inhibitor indomethacin blocked the increase in PGE₂ concentrations in cerebrospinal fluid after an injection of preformed pyrogenic factor (PFPF) and ET-1, with no antipyretic effect.^84,85 These results suggest that there is not always a direct correlation between the febrile response and increases in PGE₂ in the central nervous system. Despite the unquestionable evidence of the participation of these prostanoids in fever, other systems and/or central mediators may also contribute to the febrile response.

Corticotropin-Releasing Factor and Urocortin

CRF and urocortins (Ucns) comprise a family of peptides that are mainly involved in the stress response, but they are also important mediators of febrile responses. CRH, mainly synthesized in parvocellular cells in the paraventricular nucleus, is considered the most important hormone that is involved in the release of adrenocorticotropic hormone from the anterior pituitary gland. Once released, CRF acts on 2 types of G-protein-coupled receptors, CRF₁ and CRF₂.⁸⁶ CRF binds with high affinity to CRF₁ receptors but also has some affinity for CRF₂ receptors. Conversely, Ucn₁ binds to both receptors with similar affinity, whereas other Ucn family members (Ucn₂ and Ucn₃) preferentially act on CRF₂ receptors.^86,87 CRF₁ receptors are widely distributed throughout the brain, whereas CRF₂ receptors are preferentially expressed in the dorsal raphe, lateral septum, and hypothalamus.⁸⁸

Early studies indicated that CRF injections in the third ventricle in rabbits reduced the febrile response that was induced by leukocyte pyrogen.⁸⁹ According to a study by Opp et al.,⁹⁰ CRF injections in rabbits attenuated the febrile response that was induced by IL-1β, suggesting an antipyretic action of this peptide. However, in the same study, the authors showed that intracerebroventricular injections of CRF increased brain temperature.⁹⁰ Intracerebroventricular injections of the CRF receptor antagonist α-helical CRF_9-41 reduced the febrile response that was induced by an intracerebroventricular injection of IL-1β but not by peripherally administered IL-1β, suggesting that the action of IL-1β is mediated by this CRF only when the cytokine is present in the central nervous system.⁹¹ Another rat study showed that intracerebroventricular injections of CRF significantly increased resting oxygen consumption and colonic temperature.⁹² The increase in body temperature that was induced by CRF injections in the central nervous system in rats was further confirmed in later studies.^85,93-95 Subsequently, another study showed that both α-helical CRF_9-41 and an antibody against CRF inhibited the increase in body temperature and BAT activity that was induced by an intracerebroventricular injection of IL-1β in rats but not fever that was induced by intracerebroventricular injections of TNF-α or PGE₂ in rats.⁹⁶ Confirming these results, the non-peptide CRF receptor antagonist CP 154,526 also blocked the febrile response that was induced by IL-1β in rats.⁹⁷ Still unknown, however, is whether these different results are attributable to differential roles of CRF in different species (e.g., rats and rabbits) or whether CRF is involved in the febrile response that is induced only after the generation of IL-1β in the brain because all of the rat studies used this procedure.

The increase in body temperature that is induced by CRF occurs concomitantly with peripheral vasoconstriction, suggesting the activation of heat conservation mechanisms. Furthermore, we showed that the febrile response that was induced by CRF was also reduced by antalarmin, a CRF₁ receptor antagonist, suggesting the involvement of this receptor in the febrile response that is induced by this peptide.⁹⁴

Unexpectedly, CRF does not seem to be involved in the febrile response that is observed in zymosan-induced arthritis in rats,⁹⁸ although IL-1β is believed to be one of the most important endogenous pyrogens that is involved in the response that is induced by this TLR agonist.^98,99 However, Milton et al.¹⁰⁰ showed that the febrile response that was induced by Poly I:C in rabbits was reduced by peripheral injections of a CRF antagonist or anti-CRH antibody but not by central injections. Altogether, these results suggest that CRF release is important for the febrile response that is induced by IL-1β and likely other exogenous pyrogens, such as Poly I:C, but still unknown is why some febrile responses that clearly involve the release of IL-1β are not modified by CRF receptor antagonists.

The febrile response that is induced by CRF also does not seem to involve subsequent PG synthesis because it was not blocked by the nonselective COX inhibitors ibuprofen or indomethacin or the COX₂ inhibitor celecoxib.^94,101 This peptide also does not seem to be released after PGE₂ synthesis, in which the febrile response that was induced by an intracerebroventricular injection of PGE₂ in rabbits was unaffected by the CRF receptor antagonist α-helical CRF_9-41.⁹¹

The other member of this peptide family, Ucn₁, increases body temperature in rats but not concomitantly with the activation of heat-conservation mechanisms (such as vasoconstriction characterized by a reduction of tail skin temperature in rats).⁹⁴ Ucn₁ induced a hyperthermic response that was characterized by an elevation in tail skin temperature, which seemed to occur independently of the synthesis or release of any mediators.⁹⁴ Astressin, a CRF₂ receptor antagonist, blocked Ucn₁-induced hyperthermia.⁹⁴ In another study, Ucn₁ induced hyperthermia, which was blocked by α-helical CRF_9-41 and antalarmin but not astressin 2B, suggesting the involvement of specifically CRF₁ receptors.¹⁰² In contrast, Ucn₂- and Ucn₃-induced increases in body temperature were not modified by α-helical CRF_9-41 or antalarmin, whereas astressin 2B was effective, suggesting that CRF₂ receptors are involved in this response.¹⁰² The reason for this discrepancy is still unclear.

Endothelins

Endothelins are a family of peptides that consist of 3 isoforms and are potent vasoconstrictors of blood vessels in mammals.¹⁰³ ET-1 was isolated and sequenced in 1988 from the supernatant of aortic endothelial cells. ET-2 and ET-3 were discovered after isolation of the ET-1 gene.^103,104 ET-1 is formed as a pro-peptide (pre-proendothelin-1), which is then cleaved into big-endothelin-1, the circulating functionally inactive form of this peptide.^105,106 Conversion enzymes, kinases, and metalloproteases cleave big-endothelin into ET-1, which contains 21 amino acids. The expression of endothelin receptors (ET_A and ET_B) is regulated in parallel with peptide expression. Both receptors are G-protein-coupled receptors.^107,108 ET_A receptors have greater affinity for ET-1 and ET-2 and lower affinity for ET-3, whereas ET_B receptors have similar affinity for all isoforms.¹⁰⁹

In 1998, a study showed that central injections of ET-1 increased body temperature in rats.¹¹⁰ A subsequent study showed that ET-1 injections induced a proper febrile response, in which heat-conservation mechanisms (e.g.,, tail vasoconstriction) were activated after the injections.⁸⁴ Furthermore, ET-1 injections in the AH/POA but not intravenous injections also induced fever,^84,85 suggesting that ET-1 that is synthesized in the brain is important for the febrile response.

Several stimuli, including inflammatory stimuli, can induce the production of ET-1. Such inflammatory stimuli include endotoxins, IL-1β, TNF-α, and transforming growth factor-β.¹¹¹ ET-1 is primarily produced by endothelial cells, but other cell types (e.g., smooth muscle cells, macrophages, Kupffer cells, and nervous cells) are also potential sources of this peptide.¹¹¹ During the LPS-induced febrile response, the levels of immunoreactive big-endothelin-1 become undetectable.⁸⁴ Additionally, the levels of ET-1 in cerebrospinal fluid, which were undetectable in vehicle-treated animals, reached approximately 22 fmol/ml after an injection of LPS.⁸⁴

The selective ET_B receptor antagonist BQ788, when injected in the third ventricle in rats, blocked the LPS-induced febrile response, whereas the ET_A receptor antagonist BQ123 was ineffective, suggesting that ET-1 acted on ET_B receptors.¹¹⁰ Surprisingly, the febrile response that was induced by central injections of IL-1β and TNF-α (2 of the main cytokines that are involved in the febrile response that is induced by LPS) was unaffected by BQ788 treatment.¹¹⁰ Intracerebroventricular injections of ET-1 increased PGE₂ levels in the hypothalamus.^85,112 Both the febrile response and the increase in PG levels that were induced by ET-1 in the hypothalamus were reduced by an IL-1 receptor antagonist, suggesting that IL-1β is released after ET-1 to induce fever.⁸⁵ However, although treatment with the nonselective COX inhibitor indomethacin reduced PG levels, the febrile response that was induced by ET-1 remained unaffected.^{110, 112} Altogether, these results confirm that ET-1 is one of the central mediators of the febrile response that is induced by LPS. ET-1 appears to act on ET_B receptors, activating the febrile response that depends on IL-1β synthesis and occurs independently of PG synthesis.

Although PGs do not appear to be involved in the febrile response that is induced by ET-1, other mediators may participate in this response. For example, the μ opioid receptor-selective antagonist CTAP reversed the febrile response that was induced by ET-1, suggesting that EOps that act on μ-opioid receptors may be released after ET-1 injection to induce fever.⁹⁵ Other central mediators, such as the chemokine CCL3/MIP-1α (discussed below), also do not appear to be involved downstream of the release of ET-1.¹⁷

Recently, Kanashiro et al.⁹⁸ showed that ET-1, in contrast to LPS, did not participate in the febrile response that was observed in zymosan-induced arthritis. Bastos-Pereira et al.⁷² showed that ET-1 did not participate in the febrile response that was induced by Poly I:C. Therefore, although ET-1 is an important central mediator of the febrile response that is induced by LPS from Gram-negative bacteria, it does not participate in the febrile response that is induced by any other stimulus.

Chemokines

Chemokines are a family of structurally related proteins (8-15 kDa) that play a pivotal role in inflammation, sharing the ability to activate specific subsets of leukocytes and induce cell migration.^113,114 Chemokines have been subdivided into 4 subfamilies by considering the position of either one or 2 conserved cysteine (C) residues that are located near the N-terminus of the protein.^114,115 The CC family (β-chemokines), CXC family (α-chemokines), and CX3C family have 4 cysteine residues, with zero, one, and 3 amino acids, respectively, that separate the first 2 cysteine residues. C chemokines have only the second and fourth cysteine residues that are found in other chemokines. The C family includes lymphotactin/XCL1. The CC family includes CCL3/MIP-1α, CCL4/MIP-1β, CCL2/monocyte chemotactic protein (MCP)-1, CCL8/MCP-2, CCL7/MCP-3, CCL5/RANTES,¹¹⁶ and CCL11/eotaxin, among others. The CXC family includes CXCL^-8/IL-8, CXCL1/CINC-1, CXCL^-2/MIP-2, and CXCL^-7/neutrophil activating peptide-2, among others. The CX3C family includes CX3CL1/fraktalkine.¹¹⁷ CC chemokines, such as CCL3/MIP-1α, are attractants of mononuclear cells/macrophages and eosinophils. The C family is selective for lymphocytes, and CX3C chemokines act on diverse cell types, including monocytes, T cells, and natural killer cells.^118,119

The role of chemokines in the febrile response was first demonstrated by the ability of MIP-1 (a CC family member) to induce fever in rabbits and rats.^82,83 Sherry et al. (1988) showed that MIP-1 had 2 components: MIP-1α and MIP-1β, referred to as CCL3 and CCL4.¹²⁰ Although structurally similar, CCL3 and CCL4 have different signaling capabilities and different inflammatory activities in vitro and in vivo.¹²¹ Both CCL3 and CCL4 induced febrile responses that were insensitive to the action of indomethacin and dexamethasone in rats,^83,122,123 suggesting a direct action in the AH/POA, regardless of PG synthesis. CCL3 increased PGE₂ in cerebrospinal fluid, which was blocked by indomethacin, but the febrile response remained intact.¹²³ However, CCL3 was also shown to not belong to endogenous mediators that orchestrate the febrile response that is induced by endotoxin. For example, an anti-CCL3 antibody did not alter endotoxin-induced fever but was blocked by the CRF antagonist α-helical CRF_9-41.¹⁷ Fever that is induced by CCL4 also occurs independently of PG synthesis, but unlike what was found for CCL3, an anti-CCL4 antibody that was injected in the AH/APO reduced LPS-induced fever, indicating that this chemokine is an endogenous mediator of the pyrogenic cascade that occurs during LPS-induced fever in rats.^124,125 One possibility is that CCL3 participates in the febrile response that is induced by living infectious agents or pathogen-associated molecular patterns that are different from LPS. For example, fever that was induced by replicating Staphylococcus aureus was inhibited by central injections of anti-CCL3 antibodies (Fig. 1). Interestingly, increases in CCL3 expression after S. aureus infection in the mouse brain and human osteoblasts cultures were reported.^126,127

Figure 1.

Effect of anti-CCL3 antibody on the febrile response evoked by S. aureus. The anti-CCL3 antibody was injected intracerebroventricularly at a dose of 10 ng, both 15 min and 1 h before an intraperitoneal challenge (time 0 h) with 10¹⁰ CFU of S. aureus. Body temperature (Tb) was measured with a battery-operated biotelemetry transmitter in male Wistar rats, weighing 200 g. The data were analyzed using 2-way repeated-measures analysis of variance, followed by the Bonferroni post hoc test. The data are expressed as the mean ± SEM of the change in Tb (ΔTb, °C) for each treatment, and n indicates the number of animals in each group. *p < 0.05, compared with saline (SAL)/S. aureus group. (Martins JM , Soares DM, Souza GEP, unpublished results).

CCL5/RANTES, another member of the CC family, acts on CC chemokine receptor 1 (CCR1), CCR3, and CCR5. CCL5/RANTES is involved in LPS-induced fever in rats.⁸⁰ Injections of this chemokine in the AH/POA in rats promoted an integrated febrile response that was accompanied by concomitant peripheral vasoconstriction. However, in contrast to CCL3 and CCL4, fever that was induced by CCL5 was accompanied by an increase in PGE₂ concentration in cerebrospinal fluid, and both the febrile response and elevation of prostanoid levels in cerebrospinal fluid were prevented by pretreatment with both nonselective and selective cyclooxygenase inhibitors, including indomethacin.^80,128 Machado et al..⁸⁰ found that injections of Met-RANTES, a CCR1 and CCR5 receptors antagonist, inhibited LPS-induced fever indicating that these receptors may also be involved in this response. CCL5 also plays a major role in the febrile response that is induced by S. aureus infection.⁷³

Boddeke et al. showed that cells from the POA that were responsive to CCL5 did not respond to CCL2/MCP-1 and vice versa.¹²⁹ This implies that CCL2 activates CCR2, whereas CCL5 stimulates CCR1 and/or CCR5 in distinct subpopulations of microglia. In this context, we investigated whether CCL2, similarly to CCL5, is able to induce fever when injected in rats. An intracerebroventricular injection of this chemokine did not alter basal temperature, although it induced the migration of immune cells to the peritoneal cavity in another group of rats after intraperitoneal CCL2 injection (Fig. 2). One speculation is that neither CCR2 nor the CCR2 agonist CCL2 is involved in the central mediation of fever.

Figure 2.

CCL2 does not increase rectal temperature but induces cell migration in rats. (A) Male Wistar rats received an intracerebroventricular injection of 1 μl of saline (SAL) or CCL2 (5, 50, 250, and 500 ng), and body temperature (Tb) was measured by a tele-thermistor probe coupled to a thermometer for 6 h. (B) Male Wistar rats were euthanized 6 h after an intraperitoneal injection of 50 or 500 μg CCL2, and the peritoneal content was collected. Cells from the peritoneal cavity were collected by an injection of 10 ml phosphate-buffered saline. The abdomens were gently massaged, and a blood-free cell suspension was carefully withdrawn with a syringe. The abdominal lavage was placed in plastic tubes, and total cell counts were immediately performed in a Neubauer chamber. The data were analyzed using one way or 2-way repeated-measures analysis of variance, followed by the Bonferroni post hoc test. The data are expressed as the mean ± SEM of the changes in body temperature (A, ΔTb, °C) or number of cells per mm³ (B), and n indicates the number of animals in each group. *p < 0.05, compared with the control group (Soares DM and Souza GEP, unpublished results).

Corroborating these data, Tavares and Miñano.¹³⁰ showed that microinjections of a neutralizing antibody against CCR5 in the POA did not affect the increase in body temperature that was induced by CCL4. However, pretreatment with the same dose of anti-CCR5 antibody suppressed the febrile response that was induced by CCL5 that was administered in the same region. Thus, these authors demonstrated that hypothalamic CCR5 is functionally involved in fever that is induced by CCL5 but not CCL4. The latter has been suggested to stimulate CCR1. Therefore, although CCL4 and CCL5 are structurally and functionally similar, each has distinct characteristics that enable them to independently regulate specific aspects of the inflammatory response in the host.

Recently, the pyrogenic activity of another member of the CC chemokine family was demonstrated. CCL22 is a selective, high-affinity ligand of CCR4. In 2011, Osborn et al.⁸¹ found that an injection of CCL22 in the AH/POA increased core body temperature in mice, and this effect was mediated by the stimulation of BAT thermogenesis also in a PGE₂-dependent pathway. In parallel, these authors also found that transcripts that encoded CCL22 and CCR4 were present in the AH/POA. The authors concluded that neurons in the AH/POA express functional CCR4, which in turn respond to CCL22 with an increase in thermogenesis, thus providing an important link between neuroinflammation and thermoregulation.⁸¹

Altogether, these studies suggest that the CC chemokine family and its receptors are extensively involved in the genesis of the febrile response, activating both PG-dependent and -independent mechanisms. Chemokines are well known to be promiscuous in the relationship to the receptors they activate. Therefore, the activation of PG-dependent and -independent febrile responses may be related to these receptors. However, unless specific receptor agonists and antagonists are available and tested, this issue will remain unsolved.

Some members of the CXC chemokine family also appear to exert pyrogenic effects. Rothwell et al.¹³¹ were the first to demonstrate the ability of CXCL8/IL-8 to induce fever. These authors also showed that the febrile response that was induced by this chemokine was not reduced by the nonselective COX inhibitor ibuprofen. Subsequently, we found that intracerebroventricular CXCL8 injections dose-dependently increased rectal temperature in rats.¹² Indomethacin, at doses that reduced fever that was induced by an endotoxin from E. coli or by IL-1β, did not reduce the pyrogenic response to CXCL8, suggesting that the febrile response that is induced by this chemokine is not mediated by PGs in that species. However, differently from rats, indomethacin blocked the febrile response that was induced by central or peripheral administration of CXCL8 in rabbits.¹³² In humans, both CXCL8 and its 2 receptors, CXC receptor 1 (CXCR1) and CXCR2, are expressed, whereas in rats, no CXCL8 ortholog is expressed, but CXCR2 is found. However, rats express CXCR2 and relevant CXCR2 ligands, such as CXCL1/CINC-1 and CXCL2 (MIP-2), which are functional analogs of CXCL8 with regard to neutrophil recruitment and activation.¹³³ Our group evaluated the pyrogenic activity of CXCL1. CXCL1 injection in the AH/POA induced a febrile response that was effectively blocked by indomethacin and ibuprofen, suggesting that the febrile response that is induced by this chemokine depends on PGE₂ generation.¹⁴ Supposing that CXCL8 and CXCL1 share the same receptor for signal transduction in rats, one possibility is that these chemokines share the same intracellular pathways upon binding to these receptors, culminating in the synthesis or release of the same array of mediators. This is true with regard to the hyperalgesic effects of both chemokines in rats. Indomethacin does not affect this response, whereas the β-adrenoceptor antagonist atenolol markedly attenuates it.^134,135 However, with regard to fever, there is a clear difference between the febrile responses that are induced by CXCL1 and CXCL8.

In rats, CXCR2 has another typical ligand. CXCL2/MIP-2 appears to play an important peripheral role in regulating the mechanisms that initiate fever in both immunocompetent and leukopenic rats.¹³⁶ In this study, the febrile response that was induced by LPS was accompanied by a pronounced increase in serum CXCL2 concentrations. Systemic administration of anti-rat CXCL2 antibody significantly attenuated the early phase of LPS-induced fever. Unfortunately, there are no data on the involvement of PGs in the febrile response that is induced by CXCL2.

Table 1 summarizes the findings of the studies discussed above and the importance of chemokines as central mediators of fever. Studies that utilized different stimuli to induce the febrile response suggest the key participation of some chemokines among fever mediators in the central nervous system. For example, both mRNA and protein levels of such chemokines as CXCL10, CCL2, CCL5, CCL3, and CXCL1 increased in mouse brains that were infected with West Nile Virus,¹³⁷ and these animals presented intense fever. Moreover, much evidence suggests a role for chemokines as biomarkers in different fever-related diseases, including Kawasaki disease.¹³⁸ (i.e., an acute febrile vasculitis) and dengue hemorrhagic fever.¹³⁹ CXCL12 may also be involved in the febrile response that is induced by the human immunodeficiency virus-1 envelope glycoprotein gp120.¹⁴⁰ In Ebola hemorrhagic fever patients, CCL2, CCL3, CXCL1, and CXCL8 levels were all elevated when the prognosis indicated a fatal outcome.¹⁴¹ Much is already known, but further studies are clearly needed to improve the knowledge of the effects of chemokines on fever. Novel information regarding mediation of the febrile response will assist in improving disease diagnosis and providing safer treatments of fever symptoms that are associated with these diseases.

Table 1.

Chemokines and their putative receptors involved in mediation of febrile responses. Summary of the chemokines and the chemokine receptors among mediators involved on the mediation of febrile response. The table shows studies performed with different species, different sites of testing of chemokine receptor agonists or antagonists, if there is involvement or not of PGs, and if the agonist action on brain results in physiological consequences which can be characteristic of integrated febrile response.

Chemokine	Receptor involved	Species	Site of testing	Involvement of PGs	Physiological consequences	Contribution
CCL3	CCR1, CCR5	Rats	Lateral ventricle, AH/POA	No	Fever, HLI decrease	Refs. 17, 123
CCL4	CCCR1		AH/POA	No	Fever	Ref. 130
CCL5	CCR1, CCR5	Rats	AH/POA AH/POA	Yes Yes	HLI decrease	Refs. 80, 128
CCL22	CCR4	Mice	AH/POA	Yes	↑BAT activation ↓ respiratory exchange	Ref. 81
CXCL8	CXCR1, CXCR2	Rabitts Rats	Intravenous, i.c.v.	Yes No	Fever	Refs. 12, 132
CXCL1	CXCR2	Rats	i.c.v., AH/POA	Yes		Ref. 8
CXCL2	CXCR2	Rats	intravenous	—	Initiates LPS fever?	Ref. 14
CXCL10		Humans	Not investigated	?	↑ in periodic fever, swine fever virus, Kawasaki disease, dengue	Refs. 138, 139
CXCL12	CXCR4	Rats	Antagonist at AH/POA ↓HIV fever	?		Ref. 140
CCL2	CCR3 ? CCR4 ?		i.c.v.	—	No fever	Fig. 2, present review.

Abbreviations: AH/POA, anterior hypothalamus/pre-optic area; BAT, brown adipose tissue; HLI, heat loss index, i.c.v., intracerebroventricular; PG, prostaglandins.

Endogenous opioids

Endogenous opioids act on 3 distinct receptors (μ, κ, and δ),^142-144 and they are present in the AH/POA.¹⁴⁵ μ-Opioid receptors are also found in the ventromedial POA,¹⁴⁵ an area that is also activated during the febrile response that is induced by LPS and PGE₂.^146,147

The involvement of EOps in thermoregulation and fever has been extensively studied. The activation of μ-opioid receptors by selective agonists produces hyperthermia,^148-151 whereas hypothermia is observed after the activation of κ-opioid receptors.^151,152 With regard to the role of δ-opioid receptors in thermoregulation, the available information is contradictory.^151,153,154 Recent studies have suggested that the activation of these receptors causes hypothermia.^155,156 Conversely, EOps do not appear to be involved in maintaining normal body temperature. Central injections of CTAP, a μ-opioid receptor antagonist, or nor-binaltorphimine (nor-BNI), a κ-opioid receptor antagonist, did not modify normal body temperature, at least at low doses.^95,157 At higher doses, there seems to be a tonic balance between μ and κ opioid receptors, in which CTAP induces hypothermia and nor-BNI induces hyperthermia, and these effects can be blocked by nor-BNI and CTAP, respectively.^157,158

In 2008, Fraga et al.⁹⁵ showed that intracerebroventricular or intra-hypothalamus injections of the μ-opioid agonist morphine increased body temperature concomitantly with peripheral vasoconstriction, suggesting the activation of heat-conservation mechanisms and indicating that this response is actually a febrile response. Additionally, both nonspecific and μ-opioid receptor-specific agonists increased the activity of hypothalamic cold-sensitive neurons and reduced the activity of warm-sensitive neurons.^159,160

Previous studies have shown that EOps are involved in the febrile response that is induced by exogenous pyrogens. During the LPS-induced febrile response, μ-opioid receptors appear to be activated. Naloxone and CTAP, nonspecific and μ-opioid receptor-specific antagonists, respectively, blocked the febrile response.^161,162 Additionally, μ-opioid receptor knockout mice did not develop an LPS-induced febrile response.¹⁶³ These results suggest that μ-opioid receptors are involved in the LPS-induced febrile response. In addition to their role in the central nervous system, peripherally generated opioids may play a role in the febrile response that is induced by LPS in guinea pigs.¹⁶⁴

The administration of selective opioid receptor antagonists reduced the febrile response that was induced by TNF-α, IL-6, and CCL4/MIP-1β, but they did not modify the febrile response that was induced by IL-1β in guinea pigs and rats.^150,161,162 Other studies showed that IL-1-induced fever was blocked by buprenorphine,¹⁶⁵ which has been reported to have high affinity for μ-, κ-, and δ-opioid receptors, with K_i values in the nanomolar range.¹⁶⁶ The initial phase of interferon-α-induced fever was also reduced by the nonspecific opioid antagonist naloxone in rats.¹⁶⁷ The febrile response that was induced by PGF_2α, CRF, and ET-1 but not fever that was induced by PGE₂ was also attenuated by μ-opioid receptor blockade.⁹⁵ Furthermore, indomethacin did not change the febrile response that was induced by morphine, and μ-opioid receptor blockade by CTAP did not modify the increase in PGE₂ levels in cerebrospinal fluid or increase in COX-2 expression in the hypothalamus after LPS injection, suggesting that the release of endogenous opioids occurs concomitantly with or after PG synthesis.⁹⁵

Endogenous opioids are not only involved in the febrile response that is induced by LPS from Gram-negative bacteria; they are also involved in fever that is induced by other exogenous pyrogens of viral origin, such as Poly I:C.⁷² and glycoprotein gp-120 from human immunodeficiency virus-1.¹⁶⁸ To our knowledge, no data are available on the participation of opioids in fever of fungal origin.

Substance P

Substance P is a polypeptide that contains 11 amino acids (Arg-Pro-Lys-Pro-Gln-Gln-Phe-Phe-Gli-Leu-Met-NH₂) and is more frequently found in primary afferent A fibers, C fibers, and capsaicin-sensitive fibers.¹⁶⁹ However, the presence of substance P in the brain has also been demonstrated, including in rats, humans, and monkeys.¹⁷⁰ Some rodent studies showed that substance P can also be formed by macrophages, eosinophils, lymphocytes, and dendritic cells, among others.^171-173 Its effects are mediated almost exclusively by metabotropic neurokinin-1 (NK₁) receptors. NK₁ receptors are expressed in several brain structures, including the putamen, caudate nucleus, and hypothalamus, and in the dorsal root ganglion, intestinal intrinsic neurons, macrophages, neutrophils, lymphocytes, and mast cells.^172,174-178 Additionally, some in vitro studies have shown that substance P induces the release of cytokines from rodent macrophages and granulocytes, which are key elements of the amplification of inflammatory and febrile responses.^179-181 Lipopolysaccharide can also increase NK₁ receptor expression in some of these cells.¹⁷¹

There is high expression of substance P receptors in the rat hypothalamus, a region that is critically involved in temperature control and fever responses.¹⁸² Substance P and its precursor preprotachykinin A have also been found in the hypothalamus in primates and rats.^170,183 Unclear, however, is which cell type is responsible for the formation and release of substance P. Also unclear is where this neuropeptide, once released, acts in the hypothalamus. However, the presence of preprotachykinin A and NK₁ receptors in this region is an important observation.

In 1982, Rewerski et al.¹⁸⁴ reported that substance P injections in both the hypothalamus and lateral ventricle in rats did not alter body temperature in animals. However, a few years later, other studies showed that substance P actually induces fever and may be involved in the LPS-induced febrile response.^185,186 Additionally, an intrapreoptic injection of substance P induced prostaglandin-dependent fever.¹⁸⁵ These authors' results concerning the firing rate of warm-sensitive neurons in the anterior hypothalamus in rats did not support their results, since substance P increased the firing rate of warm-sensitive neurons while IL-1β decreased this activity. The discrepancy among these studies may be related to the metabolism of bolus injections of substance P by endopeptidases, including angiotensin-converting enzyme. We also observed variations in the response in animals after substance P injections, and the consistent induction of fever was only observed when the animals were treated with the angiotensin-converting enzyme inhibitor captopril.¹⁸⁷

The ability of LPS to induce the release of substance P in the hypothalamus has not been convincingly demonstrated, but LPS has been shown to induce the synthesis and release of substance P both peripherally.^188,189 and in the spinal cord.¹⁹⁰

Blatteis et al.¹⁸⁵ showed that an analog of substance P with broad receptor antagonist activity, when injected in the AH/POA, blocked the febrile response that was induced by LPS in guinea pigs. The participation of substance P in the LPS-induced febrile response was further confirmed by Szelenyi et al.¹⁸⁶ In this study, the febrile response that was induced by substance P was observed in restrained female rats that had been maintained at an ambient temperature of 4°C for 3 weeks, which is a condition that is far from normal. The LPS-induced febrile response was blocked by an intracerebroventricular injection of the same substance P receptor antagonist.¹⁸⁶ These studies indicate that centrally released substance P may be important for the febrile response. More recently, our group confirmed the participation of substance P and an NK₁ receptor antagonist in the LPS-induced febrile response using a more specific antagonist, SR140333. In this study, a peripheral injection of SR140333, which does not cross the blood-brain barrier, did not change the LPS-induced febrile response, whereas intracerebroventricular administration reduced this response.¹⁸⁷ However, substance P does not appear to be involved in the febrile response that is induced by IL-1β or CCL^-3, 2 important pyrogens that induce fever through PG-dependent and -independent mechanisms, respectively.¹⁸⁷ Therefore, substance P might be released after other mediators are released during LPS-induced fever, but further studies are necessary to clarify this issue. Interestingly, in contrast to LPS-induced fever, substance P does not appear to be involved in the febrile response that is induced by the viral mimetic Poly I:C.⁷²

Antipyretics With Mechanisms Unrelated to Cyclooxygenase Inhibition

The studies presented above suggest that during fever, 2 pathways run in parallel. One of the pathways is indomethacin-sensitive and requires peripheral TNF-α, IL-1β, IL-6, and some chemokines (e.g.,, CCL5, CXCL1, and CCL22) that trigger the synthesis of PGE₂, PGF_2α, and CRF in the brain. The other pathway is PG-independent, instead depending on the activation of a central endothelinergic system via ET_B receptors that is positioned downstream of PFPF and CRF or is activated through the release of chemokines, such as CCL3, CCL4, and CXCL8 (Fig. 3). The ability of IL-1β, IL-6, and PGF_2α to induce the central release of CRF.^85,92 offers the possibility of the integration of both pathways.

Figure 3.

Proposed main central pathways that are activated during fever. The left side shows the prostaglandin-dependent pathways that are activated by TNF-α, CCL5, CXCL1, IL-1β, and IL-6. The right side shows the prostaglandin-independent pathways that are activated by PFPF, CXCL8, CCL3, and CCL4. CRF release may be a key mediator of the integration of both pathways directly or through the release of ET-1 and endogenous opioids (EOp). Dashed arrows represent pathways that have not been thoroughly investigated. Substance P was omitted because its role in these pathways has not yet been elucidated.

Most antipyretics are known to inhibit COX-derived PGE₂ to promote its actions, but uncertain is how antipyretics act on the PG-independent pathway to reduce fever that is induced by these mediators. The antipyretic effect of classic NSAIDs, such as aspirin, ibuprofen, indomethacin, and coxibs (COX-2 inhibitors), has been attributed to COX-1 and/or COX-2 inhibition and the subsequent reduction of PG generation.^191,192 Nevertheless, the doses of these drugs that are required to block fever that is induced by LPS appear to be higher than those that are necessary to block PG synthesis. This observation seems to support the notion that the antipyretic effects of such drugs are produced through COX-independent mechanisms.^71,101 Thus, ibuprofen, aspirin, and salicylate at higher concentrations than those that are required to block COX can inhibit the activation and translocation of transcription factors, such as nuclear factor-κB, and consequently down regulate proinflammatory cytokines.^193-195 Ibuprofen was shown to attenuate (i) the recruitment/activation of myeloperoxidase-expressing cells in the liver, (ii) the increase in hepatic COX-2 protein, and (iii) the increase in serum TNF-α in mice that were challenged with the anti-Fas antibody Jo2.¹⁹⁶ Moreover, the release of arginine vasopressin from the ventral septal area appears to be involved in the antipyretic mechanism of action of indomethacin and salicylates, in which a V₁ receptor antagonist that was injected in the ventral septal area blocked antipyresis that was promoted by these NSAIDs.^197-199 In rats, indomethacin antipyresis is accompanied by an increase in arginine vasopressin levels in ventral septal area perfusion fluid.^197,199,200 Arginine vasopressin has been reported to induce antipyresis by stimulating septal neurons via V₁ receptors rather than by V₂ receptors.²⁰¹ This stimulation is transmitted via septofugal fibers to hypothalamic thermoregulatory structures, leading to the neutralization of neuronal activity that is triggered by endogenous pyrogens.^201,202

Our group also found that ibuprofen not only reduced fever that was induced by PGE₂-dependent endogenous pyrogens (i.e., IL-1α, TNF-α, and IL-6), similarly to indomethacin, but also reduced fever that was induced by PG-independent endogenous pyrogens (i.e., ET-1 and PFPF). Additionally, similar to indomethacin, antipyresis that was promoted by ibuprofen was antagonized by a V₁ receptor antagonist.¹⁰¹ Thus, one possibility is that the effectiveness of ibuprofen in reducing fever that is induced by PFPF, ET-1,¹⁰¹ and CCL3,¹²³ all known PG-independent pyrogens, could be related to arginine vasopressin release.

Diflunisal is an antiinflammatory and analgesic salicylate derivative that is not metabolized to salicylic acid, has less of an effect than aspirin on platelet function in vivo,^203,204 and does not cross the human blood-brain barrier (it is not found in cerebrospinal fluid in humans),²⁰⁵ all of which should be related to the absence of antipyretic activity, at least in humans.

Acetaminophen, also called paracetamol, is one of the world's most frequently used drugs to treat fever and pain. Like aspirin and other NSAIDs, acetaminophen exerts analgesic and antipyretic effects through COX-2 inhibition, without affecting mPGES-1 activity or signaling cascades downstream of PGE₂.²⁰⁶ Moreover, in contrast to indomethacin and ibuprofen but similarly to dipyrone, acetaminophen does not depend on arginine vasopressin to exert its antipyretic effect.^198,199

Also important is the effect of acetaminophen on Tityus serrulatus venom (Tsv)-induced fever. Tsv induces fever in the absence of PG synthesis in the hypothalamus and cerebrospinal fluid²⁰⁷ and appears to depend on kinins (via B₁ and B₂ receptors), IL-1β, nitric oxide, and vagal neurotransmission, which might change the fire rate of thermoregulatory centers.¹¹⁶ Like dipyrone,²⁰⁷ acetaminophen also abolished PG-independent fever that was induced by ET-1 and Tsv (Fig. 4), suggesting that in addition to blocking COX-2,²⁰⁶ other mechanisms may be involved in its antipyretic effect.

Figure 4.

Effects of acetaminophen on the febrile response evoked by ET-1 and Tsv. Acetaminophen (230 mg/kg, p.o.) or vehicle (Veh; 10% ethanol in saline plus 2 µl of Tween 80) was administered 30 min before an intracerebroventricular injection of ET-1 (1 pmol, A) or by an intraperitoneal injection of Tsv (150 µg/kg, B), or the same volume of saline (SAL). Body temperature (Tb) was measured by a tele-thermistor probe coupled to a thermometer for 6 h in male Wistar rats, weighing 200 g. The data were analyzed using 2-way repeated-measures analysis of variance followed by the Bonferroni post hoc test. The data are expressed as the mean ± SEM of the change in for each treatment Tb (ΔTb, °C), and n indicates the number of animals in each group. *p < 0.05, compared with Veh/ET-1 or Veh/Tsv group (De Moraes LA, Melo MCC, Souza GEP, unpublished results).

To state that an antipyretic has additional effects may be questionable because Tsv does not induce or depend on PGE₂ synthesis to promote fever. Rats that were treated with dipyrone exhibited a dramatic reduction of PGE₂ concentrations in cerebrospinal fluid in response to LPS, with no changes in the hypothalamic content of this prostanoid.¹¹² This finding was related to the low concentration of the main antipyretic metabolite 4-methylaminoantipyrine (4-MAA), after dipyrone treatment in this brain region since animals treated (i.p.) with 4-MAA allowed a double amount of this metabolite in the hypothalamic tissue and a parallel reduction of fever and hypothalamic PGE₂ content.²⁰⁷ These findings suggest that an antipyretic drug does not necessarily need to reduce hypothalamic PGE₂ to exert its antipyretic effect, supporting the notion that brain endothelial cells are important for PGE₂ production during fever.⁴⁴ or that some antipyretic drugs act outside the brain to block the peripheral synthesis of pyrogenic mediators, including PGE₂.⁶¹ Another possibility is that the synthesis or effect of other central mediators beyond PGE₂ should also be affected by either dipyrone or paracetamol. We are currently performing studies in our laboratory to investigate these issues.

About the Authors

Dr. Glória EP Souza (right photo) published her first study on fever, co-authored with Dr. IR Pelá, in 1992 and since then has been working in the field. She was the PhD supervisor for both Dr. Aleksander R Zampronio (in 1998; left photo) and Dr. Denis M Soares (in 2008; middle photo), who are both still working in the field. Working in 3 different universities in Brazil with collaborations from other countries, the main focus of this group has been to study the mechanisms and mediators that are involved in the febrile response, with a focus on chemokines and other central mediators of fever and the effects of several antipyretic drugs.

Concluding Remarks

The studies reviewed herein reinforce the importance of PG synthesis in the central nervous system to induce fever and clearly show that other central mediators are important in this process. These different pathways are presented in Figure 3. Importantly, different stimuli may activate different pathways. The importance of these pathways may differ under different conditions, such as in fever that is observed after an injection of live E. coli.²⁰⁸ or S. aureus ⁷³ or fever of viral origin.⁷² Additionally, the ability to generate PGs in the central nervous system is not sufficient to imply the involvement of these mediators in fever. The observations that animals can present a febrile response even when PG levels in the central nervous system are low indicate that other mediators are involved in fever. The febrile response needs to be considered as an integrated response that involves the activation of systems related to heat production and conservation and the inhibition of heat loss. Unknown is whether these central mediators differentially affect these systems. Understanding the ways in which these different mediators act, the cell types or brain regions where these receptors are expressed, and the second messengers that are activated is fundamental to better understand the participation of each mediator in the febrile response, which may consequently facilitate the development of specific treatments for fever of different origins.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interests are disclosed.

Funding

We thank Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq), Fundação Araucária do Estado do Paraná, Fundação de Amparo a Pesquisa do Estado da Bahia (FAPESB), and Fundação de Amparo a Pesquisa do Estado de São Paulo (FAPESP) for financial support.