Fever and the thermal regulation of immunity: the immune system feels the heat
Abstract
Fever is a cardinal response to infection that has been conserved in warm and cold-blooded vertebrates for over 600 million years of evolution. The fever response is executed by integrated physiological and neuronal circuitry and confers a survival benefit during infection. Here, we review our current understanding of how the inflammatory cues delivered by the thermal element of fever stimulate innate and adaptive immune responses. We further highlight the unexpected multiplicity of roles of the pyrogenic cytokine interleukin-6 (IL-6), both during fever induction as well as during the mobilization of lymphocytes to the lymphoid organs that are the staging ground for immune defence. Finally, we discuss the emerging evidence that suggests the adrenergic signalling pathways associated with thermogenesis shape immune cell function.
The fever response is a hallmark of infection and inflammatory disease and has been shaped through hundreds of millions of years of natural selection. Febrile temperatures are so closely linked to the inflammatory response that heat (calor) is one of the four cardinal signs of inflammation, along with pain (dolor), redness (rubor), and swelling (tumour), as described by Celsus in ~30 BC.1 The induction of fever in endothermic (warm-blooded) animals occurs at a high metabolic cost such that a 1°C rise in body temperature requires a 10–12.5% increase in metabolic rate.2 There is mounting evidence that the increase of 1 to 4°C in core body temperature that occurs during fever is associated with improved survival and resolution of many infections. For example, the use of antipyretic drugs to diminish fever correlates with a 5% increase in mortality in human populations infected with influenza virus and negatively affects patient outcomes in the intensive care unit.3–5 Preclinical studies in rabbits infected with rinderpest virus also found an increase in mortality when fever was inhibited with the antipyretic drug acetylsalicylic acid — 70% of acetylsalicylic acid-treated animals died as a result of infection as compared with only 16% of animals with a normal febrile response.6 However, fever is not universally beneficial, particularly in cases of extreme inflammation where lowering, rather than raising body temperature has evolved as a protective mechanism.7–10 Thus, uncontrolled fever is associated with worse outcomes in patients with sepsis or neurological injuries, whereas treatments that induce hypothermia can have a clinical benefit.11,12 A challenge in ascertaining the precise value of fever in endotherms is that the antipyretics used to inhibit fever target multiple aspects of the inflammatory response besides temperature regulation.11
Ectothermic (cold-blooded) vertebrates, which last shared a common ancestor with mammals over 600 million years ago, provide an ‘experiment in nature’ in which to examine the direct impact of febrile temperatures on survival. Ectotherms as diverse as reptiles, fish, and insects raise their core temperature during infection through behavioural regulation, which leads to their seeking warmer environments (despite the risk of predation) or, in the case of bees, raising the local temperature of the hive through increased physical activity.2,13–19 Landmark studies published 40 years ago by Kluger’s laboratory showed that survival of the desert iguana Dipsosaurus dorsalis is reduced by 75% if prevented from behaviourally raising its core temperature by approximately 2°C after infection with the Gram-negative bacterium Aeromonas hydrophila.2,13,14 The heat-seeking behaviour of the desert iguana, blue-finned tuna and leech is negated by antipyretic drugs, indicating that common biochemical pathways drive fevers in ectothermic and endothermic animals.14,16,20 Surprisingly, the correlation between infection and increased temperature extends even to plants, which arose 1.5 billion years ago. For example, the temperature of the leaves from the bean plant Phaseolus vulgaris increases by around 2°C following infection with the fungus Collectotrichum lindemuthianum.21 Thermoregulation in plants occurs through mitochondrial respiration22, although it is not known whether these fever-like responses have a direct impact on clearance of infection.
The fact that fever has been retained throughout vertebrate evolution strongly argues that febrile temperatures confer a survival advantage. A long-standing mystery relates to the protective mechanisms by which fever wards off attacks by invading pathogens. One mechanism involves direct effects of febrile temperatures on the infectious potential of pathogens.23 For example, temperatures in the febrile range (40–41°C) cause a greater than 200-fold reduction in the replication rate of poliovirus in mammalian cells and increase the susceptibility of Gram-negative bacteria to serum-induced lysis.24,25 In this Review, we discuss the evidence suggesting that febrile temperatures boost the effectiveness of the immune response during infections by stimulating both the innate and adaptive arms of the immune system. We will highlight the role of the pyrogenic cytokine interleukin-6 (IL-6) in two key phases of the febrile response: firstly in driving the rise in core temperature, and secondly as a downstream effector cytokine orchestrating lymphocyte trafficking to lymphoid organs. We also describe febrile temperature as a ‘rheostat’, dialling down systemic inflammation during the return to homeostasis. Finally, we highlight new data demonstrating the overlapping signalling pathways that are involved in thermogenesis and in the regulation of the immune response. We only briefly discuss the neuronal circuitry that drives fever and the evolutionarily conserved heat shock protein (HSP) response (BOX 1), but refer the reader to recent comprehensive reviews for additional information on these topics as well as on the contributions of hypothermia to limiting inflammation.26–30
Induction of fever
The IL-6–COX2–PGE2 axis drives fever
The induction and maintenance of fever during infection involves the tightly coordinated interplay between the innate immune system and neuronal circuitry within the central and peripheral nervous systems. Immune sensing of infection begins with the binding of pathogen-associated molecular patterns (for example, lipopolysaccharide ((LPS)), viral RNA, or fungal sugars) to pathogen recognition receptors (PRRs), such as Toll-like receptors (TLs), which are expressed by innate immune cell populations, including macrophages, neutrophils and dendritic cells (DCs) (FIG. 1). Much of our current understanding of the molecular mechanisms underlying fever stems from studies in which rodents were injected with LPS, a component of Gram-negative bacterial cell walls, to model immune-induced thermoregulation. In this model, prostaglandin E2 (PGE2) produced by brain vascular endothelial cells is considered a major pyrogenic mediator of fever.31–33 This lipid effector molecule integrates input signals from pyrogenic cytokines produced in response to pathogenic stimuli, with output signals involving neurotransmitters that raise core body temperature (FIG. 1). PGE2 is also synthesized in the periphery early in this response – that is, prior to the detection of circulating cytokines. It is produced by hematopoietic cells following LPS-mediated activation of TLR4 and travels through the blood/brain barrier to initiate fever.26,30,34–38 LPS-induced fever occurs via autonomic mechanisms driven by PGE2 binding to EP3 prostaglandin receptors expressed by thermoregulatory neurons in the median preoptic nucleus within the hypothalamus.8,39–41 Endotherms elevate body temperature through the release of norepinephrine, which increases thermogenesis in brown adipose tissue and induces vasoconstriction in the extremities to reduce passive heat loss.26,27 In addition, signalling through the neurotransmitter acetylcholine stimulates the musculature to convert stored chemical energy into thermal energy and increases overall metabolic rates.2,26,42,43 Endotherms, like ectotherms, also engage in heat-seeking behavioral thermoregulation which does not require median preoptic neurons although the pathways involved are largely unknown.8–10
The recognition of damage-associated molecular patterns (DAMPs) or pathogen-associated molecular patterns (PAMPs), such as lipopolysaccharide (LPS), by Toll-like receptors (TLRs) and other pattern recognition receptors drives the activation of dendritic cells (DCs) and macrophages. These innate immune cells release prostaglandin E2 (PGE2) as well as pyrogenic cytokines (namely, interleukin-1 (IL-1) IL-6, and tumour necrosis factor (TNF)) that act systemically to induce fever. IL-6 operates downstream of IL-1 in the median preoptic nucleus region within the hypothalamus to induce the synthesis of cyclooxygenase 2 (COX2), the enzyme responsible for production of additional PGE2.64,65 PGE2 is considered the major pyrogenic mediator of fever.31–33 Receptor activator of NF-κB (RANK) expressed by astrocytes also acts via the COX2–PGE2 pathway to induce fever.47 However, it is not known whether this pathway parallels the IL-6 response or if the IL-6 and RANKL pathways converge, potentially via IL-6 regulation of RANKL expression in vascular endothelial cells in the hypothalamus. Neurons expressing PGE2 receptor 3 (EP3) trigger the sympathetic nervous system to trigger norepinephrine release, which elevates body temperature by increasing thermogenesis in brown adipose tissue as well as by inducing vasoconstriction to prevent passive heat loss.2,26,27,42,43 Additionally, acetylcholine contributes to fever by stimulating muscle myocytes to induce shivering.
LPS-induced TLR4 signalling stimulates the synthesis of pyrogenic cytokines (namely IL-1, tumour necrosis factor ((TNF)) and IL-6) at the site of infection as well as within the brain, and it is becoming clear that IL-6 is as an important mediator of fever induction.26,44–47 Notably, multiple cell types within the brain (for example, astrocytes, microglial cells and neurons) have the capacity to synthesize IL-6 in response to local inflammatory stimuli48–53. Although the direct administration of TNF, IL-1, or IL-6 into the brain leads to a febrile response, several lines of evidence point to a requisite role for IL-6 in sustaining fever. In this regard, LPS-induced fever does not occur in the presence of IL-6-specific neutralizing antibody or in IL-6-deficient mice, even though TNF and IL-1 upregulation is normal in these settings.54–58 Moreover, direct intracerebroventricular injection of IL-6, but not IL-1, restores febrile responses in IL-6-deficient mice.55 Febrile temperatures have further been implicated in a positive feedback loop during the early stages of infection. Specifically, passive elevation of the core body temperature of mice to the febrile range using whole body hyperthermia substantially augments circulating levels of TNF, IL-1, and IL-6 during LPS-induced inflammation.26,59–61 The pyrogenic role of IL-6 has recently been corroborated in patients with paediatric leukaemia, where treatment with the IL-6 receptor antagonist tocilizumab was found to reverse the high fevers that develop during T cell based-immunotherapy (specifically, following the administration of chimeric antigen receptor-expressing T cells or a CD19/CD3-bispecific antibody).62,63
Systemic or locally produced cytokines act in the brain to augment the synthesis of cyclooxygenase 2 (COX2), the enzyme responsible for oxidizing arachidonic acid to produce PGE2 (FIG. 1). For instance, IL-1 receptors that mediate COX2 induction have been identified on brain endothelial cells within the preoptic region of the hypothalamus.64,65 Although the specific cell types that upregulate COX2 in response to IL-6 remain to be identified, blood vessels in the brain reportedly express the IL-6 receptor-α subunit,53 which together with the ubiquitously expressed gp130 subunit forms the functional IL-6 receptor. Several studies have shown that cerebral COX2, PGE2, and fever are not induced during LPS-driven inflammation in IL-6-deficient mice or in the presence of IL-6-specific neutralizing antibody.66–68 Alternatively, IL-6 cannot initiate a febrile response in the absence of COX2 or PGE2, and intracerebroventricular delivery of PGE2 bypasses the requirement of IL-6 for fever induction in IL-6-deficient mice.69,70 Collectively, these observations establish that COX2 and PGE2 are crucial mediators that can operate downstream of IL-6 in the LPS-induced febrile response.
RANKL and fever induction
An open question is whether IL-6 is the direct regulator of COX2 and PGE2 induction during the febrile response or whether other intervening cytokines are involved. The latter possibility is suggested by an elegant study by Hanada et al. showing that, similarly to IL-6, the cytokine known as receptor-activator of NF-kB ligand (RANKL) converges on the COX2–EP3–PGE2 pathway leading to fever induction in the LPS-induced model of inflammation47 (FIG. 1). RANKL is best known as a regulator of bone remodelling and lymph node organogenesis.71 However, mRNA encoding RANKL is also produced in the lateral septal nucleus region of the brain that interconnects with the hypothalamus and the RANKL receptor, RANK, is found on astrocytes in the preoptic region of the hypothalamus.47 Further support for a role of this cytokine in thermoregulation is provided by findings that children with RANK mutations exhibit impaired fever responses during pneumonia.47 Although the potential interplay between IL-6 and RANKL–RANK during fever has not been explored, it is tempting to speculate that RANKL is a downstream mediator of IL-6-induced pyrogenesis based on evidence that IL-6 directly stimulates RANKL synthesis by synovial fibroblasts in mouse models of rheumatoid arthritis.72
Immune stimulation by thermal stress
One benefit widely attributed to fever is the enhancement of immune-protective mechanisms during infection. Defence against pathogens involves tight spatial and temporal regulation of the immune system, and the same pyrogenic cytokines that are produced during the induction of fever also operate locally to orchestrate immunity within infected tissues.73 Innate immune cells are the ‘first responders’, arriving within hours to directly destroy pathogens via phagocytic or cytotoxic activities. These activities limit infection until a peak adaptive immune response is generated, normally around one week later. Macrophages and DCs bridge the gap between innate and adaptive immunity by taking up pathogens in peripheral tissues and then relocating to draining lymph nodes where they drive expansion of pathogen-specific effector T cells.74,75 Crucial for this process is the co-localization of DCs and T cells near high endothelial venules (HEVs) that are the major portals for entry of blood-borne lymphocytes.74–76
Given the complexity of these immune mechanisms, it is remarkable that fever-range temperatures stimulate almost every step involved in this process, promoting both innate and adaptive immunity. The potential impact of the thermal element of fever has primarily been explored using hyperthermic temperatures within the febrile range for mammals (that is, ranging from 38–41°C; ΔT~1–4°C above baseline) in the various in vitro and in vivo studies described below. Experimental hyperthermia is a powerful approach to study the impact of febrile-range temperatures on immunity, which is otherwise difficult to discriminate during natural fever because of the attendant inflammatory programme (comprised of lipid and cytokine mediators) that regulate both fever and immunity. However, an important caveat from a physiological perspective, is that the heat conservation associated with natural fever differs fundamentally from cooling mechanisms enacted by thermoregulation following exogenous heat application.
Impact of febrile temperatures on innate immunity
Previous research using animal models of hyperthermia treatment alone, or with LPS challenge or bacterial infection, strongly supports the idea that fever-range temperatures elevate the respiratory burst typically associated with activation and bacteriolytic activity of neutrophils77,78 (FIG. 2a). Thermal stress further increases neutrophil recruitment to local sites of infection and other distant tissues61,79 (FIG. 2a) including tumours77 Neutrophil localization in peripheral tissues is due, at least in part, to heat-induced increases in circulating neutrophils which are dependent upon granulocyte colony-stimulating factor (G-CSF).80,81 G-CSF is also central in a model of radiation-induced neutropenia where fever-range whole body hyperthermia substantially increases the rate of neutrophil recovery in the blood, and augments the number hematopoietic stem cells and neutrophil progenitors within the bone marrow82 (FIG. 2a). This effect is dependent upon enhanced production of IL-17, IL-1β and IL-1α preferentially in intestinal tissue. Importantly, the precise outcome of the thermal effect depends on the heating protocol used and the geography of cell recruitment (FIG. 2a). Indeed, temperatures above the normal febrile range impair neutrophil accumulation and function.83 Moreover, Hasday and colleagues found that fever, or exposure to fever-range hyperthermia, in an LPS model increases neutrophil localization to the lung, which can have negative consequences due to inflammation-induced local tissue damage.61,84 Heat-induced neutrophil recruitment in the lung depends on the non-canonical chemotactic HSP, CXCL8 (also known as IL-8), which is expressed under the control of the heat-inducible transcription factor heat shock factor 1 (HSF1) (BOX 1).61,84,85 Neutrophil recruitment in the lung also involves a decrease in endothelial barrier integrity through a mechanism depending on p38 MAPK and ERK1–ERK2 signalling.84
(a) Fever-range temperatures drive several crucial aspects of innate immunity. Fever-range hyperthermia stimulates the release of neutrophils from the bone marrow in a granulocyte–colony-stimulating factor (G-CSF)-driven manner.80–82 Febrile-range temperatures also promote neutrophil recruitment to the lungs and other local sites of infection in a CXC-chemokine ligand 8 (CXCL8)-dependent fashion that additionally involves decreased barrier function of vessels.61,84,85 Upon arriving in the site of infection, thermal stress further elevates the respiratory burst which increases the bacteriolytic activity of neutrophils.77,78 (b) Thermal treatment improves natural killer (NK) cell cytolytic activity through induction of MHC class I polypeptide-related sequence A (MICA) expression on target cells (for example, tumour cells) as well as by inducing the clustering of the MICA counter-receptor NKG2D on the surface of NK cells.90 (c) Temperatures in the febrile range increase the ability of antigen-presenting cells to support the formation of the adaptive immune response. Heat improves the phagocytic potential of macrophages and dendritic cells (DCs) and increases their responsiveness to invading pathogens by upregulating their expression of both Toll-like receptor 2 (TLR2) and TLR4.119,120 Thermal treatment also induces the release of immunomodulatory molecules such as cytokines (for example, TNF), nitric oxide (NO) and heat shock protein 70 (HSP70). Additionally, heat increases expression of MHC class I and II molecules as well as co-stimulatory molecules (CD80 and CD86) by mature DCs and augments their CC-chemokine receptor 7 (CCR7)-dependent migration via the afferent lymphatics that serve as a conduit to draining lymph nodes.117,121–124 DCs exposed to febrile temperatures are also more efficient at cross-presenting antigens and inducing T helper 1 (Th1) cell polarization.121
The impact of heat on natural killer (NK) cells has been most extensively studied in the context of tumour immunity. It has been shown that NK cell cytotoxic activity and recruitment to tumour sites is increased by fever-range hyperthermia in vivo86–89 (FIG. 2b). This enhanced cytotoxicity depends upon heat-induced upregulation of the NKG2D ligand MICA (MHC class I polypeptide-related sequence A) on tumour cells as well as on the clustering of NKG2D receptors on the surface of NK cells.90 Elevated temperatures also decrease MHC class I expression by tumour cells while simultaneously increasing HSP70 production, and both of these responses are linked to enhanced cytotoxic potential in NK cells.91 The upregulation of HSPs in tumour cells in response to thermal stress is also likely to be involved in the enhanced cross-priming of antigen-specific cytotoxic T lymphocytes that was observed when DCs were loaded with lysate from heated melanoma cells.92
Macrophages have served as a major model for the study of febrile-range hyperthermia. Early studies demonstrated that whole-body heating (to ~39.5°C) improves bacterial clearance and also increases serum concentrations of TNF, IL-1, and IL-6 in mice challenged with LPS.59,60,93,94 The source of these cytokines was found to be the macrophages of the liver (that is, Kupffer cells) as well as macrophages in other organs. Later work by Lee et al. showed that hyperthermia induces the upregulation of HSP70 and this ‘reprogrammes’ macrophages to show sustained activation in response to LPS.95 The mechanism involves the phosphorylation of the IKK and IκB kinases, the nuclear translocation of NFκB and its binding to the Tnfpromoter.95,96 HSP70 is also required for enhancing the expression of nitric oxide and inducible nitric oxide synthase by peritoneal macrophages following exposure to fever-range temperatures together with LPS and IFNγ.97 Although HSPs are usually assumed to be intracellular, heat stress can induce HSP70 release from cells into the extracellular environment where it can act as a damage-associated molecular pattern (DAMP) to stimulate macrophages and DCs.98–100 Extracellular HSP70 and other HSPs engage multiple surface receptors, including CD91, scavenger receptor A, CD40, TLR2 or TLR4, leading to the release of nitric oxide, TNF, IL-6, IL-1β and IL-12.100–110 Of note, some investigators have paradoxically observed an anti-inflammatory role for HSPs.111–113 It is suggested that these differences result from the precise location of HSPs within macrophages: extracellular HSPs provide danger signals to enhance inflammation whereas intracellular HSP could help to suppress inflammatory signalling.114 Taken together, the data regarding innate immune cells, body temperature and HSPs reveal fascinating, yet still poorly understood, layers of interdependency between the febrile response and the more ancient HSP response.
Fever enhances DC functions
Several studies demonstrate that elevated temperatures substantially enhance the phagocytic potential of DCs, in addition to augmenting interferon-α (IFNα) production in response to viral infection (FIG. 2c).115–118 Heating of immature DCs also upregulates their expression of TLR2 and TLR4, suggesting a role for thermal signals in enhancing pathogen sensing by innate immune cells.119,120 Febrile temperatures further increase DC expression of MHC class I and class II molecules and co-stimulatory molecules, including CD80 and CD86, and can augment the secretion of the Th1 cell-polarizing cytokines IL-12 and TNF.102,117,119–123 Additional reports point to a role for febrile-range temperatures in augmenting the migration of antigen-presenting cells (APCs), such as skin Langerhans cells, to draining lymph nodes124 (FIG. 2c). These data may help to explain the fact that febrile temperatures can accelerate the swelling phase of a contact hypersensitivity reaction when heat is delivered to mice shortly after the application of the elicitation dose of a skin sensitizer, fluorescein isothiocyanate (FITC).124 The underlying mechanism directing DC migration to draining lymph nodes likely involves increased responsiveness of CC-chemokine receptor 7 (CCR7) to its ligands, which has been described for heat-treated mature DCs in chemotaxis assays in vitro.121 CCR7 senses CCL21 chemokine gradients in vivo, thereby guiding DC entry into afferent lymphatics and their subsequent migration near HEVs within draining nodes.125–128 Thus, febrile temperatures appear to regulate the CCR7–CCL21 axis in order to optimally position DCs in lymphoid organs at sites where they can present antigen to lymphocytes upon their arrival via HEVs.
Given these observations, it is not surprising that fever-range thermal stress enhances the ability of DCs to stimulate T cells as well as DC cross-presenting functions (FIG. 2c). In mixed lymphocyte reactions, applying thermal stress ex vivo to LPS-pulsed mature human monocyte-derived DCs led to enhanced proliferation of naïve CD4+ T cells and promoted their differentiation towards a Th1 cell phenotype.121 Similarly, DCs isolated from heated mice exhibit a superior ability to activate T cells.102 In studies where DCs from patients with medullary thyroid cancer were preheated prior to co-culture with T cells, the T cells showed enhanced cytotoxicity against tumour targets.119 This increased cytotoxicity of effector T cells correlated with heat-induced upregulation of both MHC class I and HSP70 expression in mature, but not immature, DCs. Together, these findings demonstrate that systemic fever-range temperatures can target different components of the innate immune system, including the HSP response, in order to enhance effector T cell responses.
Thermal mechanisms boost adaptive immunity
A crucial determinant for the generation of adaptive immunity is the high rate of lymphocyte trafficking through lymphoid organs. The entire pool of naive T cells in a mouse lymph node turn overs ~2–3 times per day as a result of T cell recirculation.75,129 This dynamic flux increases the probability that rare antigen-specific T cells (present at a frequency of only ~1 in 105–106)130,131 will receive activating signals from DCs. The entry of blood-borne T and B cells into lymph nodes and Peyer’s patches occurs preferentially at HEVs through a well-defined adhesion cascade that involves; one, L-selectin and/or α4β7 integrin initiated tethering and rolling; two, CCL21-dependent activation of CCR7 on adherent lymphocytes; three, LFA1–mediated firm arrest via binding to its endothelial counter-receptors intercellular adhesion molecule-1 (ICAM1) and ICAM2; and four, LFA1–ICAM1–2-directed transendothelial migration.74–76,132,133 As described below, we have shown that fever-range thermal stress targets multiple steps in this cascade by invoking a wide array of lymphocyte and endothelial trafficking molecules134–142 (FIG. 3a).
(a) Fever-range thermal stress supports increased adaptive immunity by targeting two distinct aspects of T cell activation in lymph nodes. Heat enhances the rate of lymphocyte trafficking across high endothelial venules (HEVs) in peripheral lymph nodes through effects on each step of the adhesion cascade. Heat treatment of lymphocytes increases the frequency of L-selectin-dependent tethering and rolling interactions.134,135,137–139 Febrile-range temperatures independently act on HEVs to enhance the transition of lymphocytes from transient rolling to stable arrest by increasing the intravascular density of CC-chemokine ligand 21 (CCL21) and intracellular adhesion molecule 1 (ICAM1).140–142 ICAM1 also supports lymphocyte crawling to inter-endothelial cell junctions as well as transendothelial migration.131,145,146 Heat also acts directly on the T cells within lymphoid organs by pre-clustering components of the immunological synapse (TCRβ and CD8) into lipid rafts. This prolongs stable contacts with APCs and increases CD8+ T cell differentiation towards an effector phenotype characterized by enhanced L-selectin downregulation, cytotoxic function, and production of interferon-γ (IFNγ).151,152 (b) Epifluorescence whole-mount confocal microscopy imaging of HEVs that are actively supporting lymphocyte trafficking in a mouse lymph node. HEVs are stained in red with PE-conjugated MECA-79 antibody that recognizes peripheral lymph node addressin (PNAD) whereas lymphocytes are labelled in green using carboxyfluorescein succinimidyl ester (CFSE).
An early indication that fever could control lymphocyte trafficking emerged from studies showing transient decreases in circulating T cells in mice or patients with cancer following elevation of core body temperatures to ~39.5°C by febrile-range whole body hyperthermia.83,137,143 Reductionist studies found that direct heat treatment of T or B cells ex vivo for 6 hours resulted in an approximately 2-fold increase in their ability to bind to HEVs in vitro or to home to lymph nodes or Peyer’s patches in vivo.134–139 Lymphocytes isolated from heated mice exhibit similar enhancement of homing properties.138 It is worthwhile noting that this represents a substantial increase above the already efficient rate of homeostatic trafficking whereby ~1 in 4 lymphocytes initiate the adhesive events that precede extravasation.75,129
Fever-range temperatures augment trafficking through a lymphocyte-autonomous mechanism by targeting the binding activity of both L-selectin (FIG. 3a) and α4β7 integrin without altering their density.134,136–139 In lymph node HEVs, fever-range hyperthermia promotes L-selectin-dependent lymphocyte rolling along vessel walls through the formation of short-lived catch-bonds with its endothelial counter-receptor, peripheral node addressin (PNAD).74,75 Febrile temperatures also enhance α4β7 integrin binding to mucosal addressin cell adhesion molecule 1 (MADCAM1) in Peyer’s patch and mesenteric lymph node HEVs.144 Direct exposure of lymphocytes to heat does not alter LFA1 affinity for its endothelial ligands.134,136 It remains an open question whether the chemokine receptor, CCR7, is responsive to febrile temperatures.
The intrinsic binding function of HEVs is also enhanced approximately 2-fold in LPS- or turpentine-induced mouse models of fever as well as during exposure of mice to fever-range whole body hyperthermia.136,137,140–142 (FIG. 3a). As in lymphocytes, maximal enhancement of HEV adhesion requires sustained temperature elevation (more than 6 hours),136,137,140–142 recapitulating the extended time-frame of physiological fever responses. Chen et al. visualized lymphocyte interactions in mouse HEVs using intravital microscopy (FIG. 3b), together with quantitative image analysis of trafficking molecules, to pinpoint the thermally responsive trafficking mechanisms in HEVs.140,141 Thermal stress does not alter the ability of HEVs to support rolling, nor does it change the intraluminal density of the prototypical rolling molecules, PNAD or MADCAM1.140,141 Instead, exposure to febrile temperatures profoundly increases the ability of HEVs to support the stable arrest of lymphocytes and this can be attributed to heat-induced increases in the intravascular density of CCL21 and ICAM1140,141 (FIG. 3a). Of note, the level of HEV adhesiveness and ICAM1 expression induced by thermal stress is equivalent to that observed in response to the potent pro-inflammatory cytokine, TNF.140 Thermal upregulation of CCL21 and ICAM1 in HEVs is consistent with the known concentration-dependent roles of these molecules in augmenting LFA1 affinity (~10,000-fold), thereby supporting stable adhesion of lymphocytes within vessel walls.145,146 Additionally, ICAM1 elevation in response to hyperthermia likely promotes LFA1-dependent transendothelial migration in HEVs and the formation of ICAM1-dense adhesive patches that guide lymphocyte diapedesis into underlying tissues.133,147,148
Once lymphocytes gain entry into lymphoid organs, there is evidence that their ability to respond to stimulatory signals is also enhanced by febrile temperatures. Direct exposure of T cells to fever-range hyperthermia increases their proliferation in response to mitogens.149,150 Furthermore, in both in vitro and in vivo models of antigen-driven T cell activation by APCs, thermally treated CD8+ T cells show greater differentiation towards an effector phenotype, with pronounced L-selectin downregulation, enhanced cytotoxic function and increased production of IFNγ151,152 (FIG. 3a). Enhanced stimulation of naive CD8+ T cells is aligned with temperature-dependent PKCβ activation, prolonged stable contacts with APCs, and transient clustering of components of the immunological synapse (TCRβ and CD8) in cholesterol-enriched microdomains,.151,152 Similar heat-induced changes in membrane fluidity and macromolecular clustering in the plasma membrane occur in CD4+ T cells which reduce the requirement for CD28 stimulation for IL-2 production.153 These findings suggest that febrile temperatures lower the threshold for T cell signalling and effector T cell differentiation by pre-associating the signalling components of the TCR complex.
IL-6 is a thermally sensitive effector of trafficking
Investigation into the mechanisms underlying thermal regulation of trafficking led to the unexpected discovery that the same pyrogenic cytokine responsible for inducing fever, namely IL-6,135,137,138 also controls both lymphocyte and endothelial adhesion.132,138–140,142 The thermal response further depends on a second soluble factor, the soluble form of the IL-6 receptor α subunit (sIL-6Rα), which acts cooperatively with IL-6 and the membrane-anchored gp130 signal transducing molecule through a well-defined mechanism termed trans-signalling138–140,154,155 (FIG. 4a). This thermally sensitive mechanism was identified in vitro and in vivo using recombinant soluble gp130138,140, which is a competitive antagonist of IL-6 trans-signalling but does not affect classical signalling involving membrane-anchored IL-6Rα.154,156
(a) Heat-dependent interelukin-6 (IL-6) trans-signalling is initiated by binding of the soluble form of the IL-6 receptor α subunit (sIL-6Rα) to both IL-6 and membrane-anchored gp130.154,155 Soluble gp130 functions as a selective antagonist of IL-6 trans-signalling and downstream activation of canonical JAK–STAT and MEK1–ERK1/ERK2 signalling pathways but does not interfere with classical signalling by membrane-anchored IL-6Rα and transmembrane gp130.156 (b) Febrile temperatures act on lymphocytes and high endothelial cells (HECs) to improve lymphocyte trafficking exclusively across high endothelial venules (HEVs) in lymph nodes. Vessel segments immediately proximal to HEVs are refractory to thermal treatment, which may reflect the lower expression of gp130 by non-specialized squamous endothelial cells that line non-HEVs.162 Left inset, fever-range temperatures act directly on lymphocytes through IL-6 trans-signalling to stimulate the MEK1–ERK1/ERK2 signalling pathway, promoting L-selectin adhesion as well as intermolecular interactions between the actin-based cytoskeleton, α-actinin, and the cytoplasmic tail of L-selectin.138 Right inset, IL-6 trans-signalling upregulates the intravascular density of ICAM1 in HEVs during heat treatment of mice, although the downstream signalling mediators remain unknown. Fibroblastic reticular cells that are in direct contact with HECs165 are a possible source of the IL-6 while proximal dendritic cells (DCs) and T cells could provide the sIL-6R138,164 required to enhance the adhesive properties of HEVs during thermal stress.
In lymphocytes, the MEK1–ERK1/ERK2 signalling pathway, but not p38 MAPK or JNK, operates downstream of IL-6–sIL-6Rα trans-signalling in response to heat.138 This promotes L-selectin interactions with actin-based cytoskeletal scaffolding elements, thereby enhancing its apparent tensile strength (FIG. 4b). IL-6-induced activation of STAT3 also occurs in lymphocytes in response to thermal stress138, although it is not known whether this contributes to lymphocyte adhesion, or delivers survival signals157,158 that aid the expansion of populations of effector lymphocytes within lymphoid organs. Consistent with the evolutionary conservation of the febrile response, L-selectin adhesion is induced by fever-range temperatures through a common IL-6 trans-signalling mechanism in animals representing four taxa of jawed vertebrates that includes endothermic mammals (for example, human, rodents, dog, cow, tiger, elephant, and rhinoceros) and birds (chicken), as well as ectothermic amphibians and fish.134,135,137–139 These observations strongly suggest that conservation of IL-6-regulated lymphocyte trafficking mechanisms over hundreds of millions of years of evolution confers a survival benefit during fever.
Gp130 ligation by IL-6–sIL-6Rα also upregulates the intravascular density of ICAM1 in HEVs during heat treatment of mice132,140 (FIG. 4b). The dual requirement for IL-6 and sIL-6Rα for ICAM1-dependent trafficking in HEVs during thermal stress is in line with the prevailing view that endothelial cells generally lack membrane-anchored IL-6Rα, and thus are refractory to IL-6 unless sIL-6Rα is available.132 STAT3 and MEK1–ERK1/ERK2 signalling have been implicated in transcriptional regulation of ICAM1132 and, thus, are potential mediators of the thermal response in HEVs. By contrast, CCL21 induction is not dependent on IL-6 trans-signalling,140 suggesting an additional molecular pathway is induced by febrile temperatures.
One of the most intriguing findings to emerge from intravital imaging relates to the tight spatial regulation of IL-6–sIL-6Rα responses in vascular beds during thermal responses. In this regard, Chen et al. showed that HEVs respond to IL-6 trans-signalling during thermal stress, but contiguous vascular segments that are not comprised of high endothelial cells (HECs) are completely refractory to thermally induced IL-6 trans-signalling (FIG. 4b).140 Similarly, non-HEVs in other organs are not responsive to febrile temperatures although heat-shock (which occurs at temperatures greater than 43°C) reportedly stimulates ICAM1 expression in normal vascular endothelium137,140–142,159,160 This restricted vascular response to physiological temperature elevation is proposed to maintain focal trafficking of lymphocytes at HEVs in lymph nodes and Peyer’s patches located throughout the body, thus maximizing their opportunity to scan pathogen-derived antigens from peripheral sites of infection.137,140,141
The mechanism maintaining spatial resolution within venular segments over distances spanning the width of a single HEC (~30 µm)161 remains to be resolved, but clues have emerged from recent transcriptional profiling of various cell subsets in lymphoid organs. HECs are distinguished from their normal endothelial cell counterparts by elevated expression of Il6st (which encodes gp130)162 that could theoretically predispose them to be highly sensitive to IL-6–sIL-6Rα in the local milieu (FIG. 4b). Although the overall nodal concentrations of IL-6–sIL-6Rα are unchanged by thermal stress,140,163 heat could theoretically induce their synthesis by discrete cell populations or lower the threshold for signalling in HECs. Fibroblastic reticular cells (FRCs) are a possible source of IL-6 during fever based on their high expression of Il6 mRNA relative to hematopoietic cells or blood endothelial cells within skin-derived lymph nodes.164 Unlike other vascular beds that are circumscribed by pericytes, HEVs are in direct contact with FRCs, and thus are optimally positioned to receive instructions from FRC-derived cytokines.74,75,165 Of particular relevance is a report that IL-6 synthesis by fibroblasts can be induced by the heat-inducible transcription factor, HSF1.166 The sIL-6Rα necessary for trans-signalling is likely provided by neighbouring leukocytes including DCs, monocytes, and/or T cells.138,164 Recent studies have shown that febrile temperatures can also act through IL-6 trans-signalling to augment the recruitment of cytotoxic CD8+ T cells across tumour-associated vessels.142 These studies are highly relevant to the use of thermal medicine as an adjuvant for cancer immunotherapy (BOX 2) and raise the possibility that fever could invoke similar mechanisms to amplify effector T cell trafficking at sites of infection.
A return to homeostasis
The immune response must be tightly regulated to avoid excessive tissue damage after infection. By extension, it makes sense that the effects of febrile temperatures on the immune system are also temporally regulated during the resolution phase of inflammation although a full picture of the underlying mechanisms is yet to emerge. One example is the rapid restoration of lymphocyte trafficking in HEVs to basal levels within 6 hours following cessation of fever-range hyperthermia.134,137,141 Normalization of HEVs is mediated by zinc-dependent metalloproteinases that cleave endothelial ICAM1 while sparing other trafficking molecules (such as PNAD),141 although it is not known if heat stimulates the catalytic activity of these enzymes. In line with a potential anti-inflammatory role of hyperthermic temperatures, heat shock (42°C for 15 min) has been found to blunt leukocyte adhesion within vessels if administered 2 day prior to the intravascular delivery of the neutrophil attractant FMLP in vivo.167
Although febrile temperatures initially increase the production of pro-inflammatory cytokines by macrophages at sites of inflammation,59–61,95,96 there is also evidence that thermal stress dampens cytokine synthesis once macrophages become activated. This sequence of events is analgous to natural fever, which often occurs after macrophages and other innate immune cells initially encounter PAMPs. In this regard, human monocyte-derived macrophages with an activated phenotype produce less TNF, IL-6, and IL-1β when exposed to febrile temperatures than heat-inexperienced cells.95,96,168–170 Heat reduces transcription of pro-inflammatory cytokines through repressive activities of HSF1, together with diminished recruitment of NF-κB to the promoter regions of cytokine-encoding genes, and also lowers cytokine mRNA stability.171–173 Thermal treatment of LPS-activated macrophages also appears to dial down inflammation by inhibiting the release of the inflammatory DAMP known as high mobility group box 1 (HMGB1), which is a ligand for TLR2 and TLR4.170,174 Inhibition of HMGB1 release prevents the subsequent activation of NF-κB, which controls the synthesis of pro-inflammatory cytokines in innate immune cells.169,170,174 The idea that heat can dampen an on-going pro-inflammatory condition in vivo has recently been tested in a murine model of collagen-induced arthritis.175 Mice exposed to fever-range hyperthermia had significantly less joint damage, correlated with a reduction in serum TNF levels and increased IL-10 production in inflamed joints. Collectively, these findings suggest that strategic temperature shifts contribute to a biochemical negative feed-back loop that protects tissues against damage from excessive cytokine release following infection.
Thermogenesis and adrenergic signalling in immunity– an emerging concept
Neural components of the thermoregulatory system continuously monitor temperature changes throughout the body and initiate integrated responses that either increase internal heat content (for example, through thermogenesis in brown adipose tissue) or increase the dissipation of heat (for example, following intense exercise).8 Given the homeostatic importance of thermoregulation, it is all the more remarkable that fever has been so long maintained in evolution, as natural thermoregulatory signals must be suppressed in order to increase body temperature. Although the examples discussed earlier in this review demonstrate that the immune system is responsive to elevated temperatures, new studies reveal that this system is also highly sensitive to the metabolic stress associated with thermogenesis. Emerging evidence strongly supports a direct role for cold stress-induced norepinephrine production and its interaction with β-adrenergic receptors on immune cells as a major mechanism for immune modulation by environmental cold stress. It is well established that norepinephrine-driven stimulation of β-adrenergic receptors is crucial for the release of additional heat from mitochondria in brown adipose tissue during cold stress to maintain a normal core body temperature.176,177 Moreover, the ubiquitous presence of β-adrenergic receptors has been observed on the surface of immune cells, and there is a growing appreciation of the functional consequences of signalling through these receptors.178–180 Even more recent studies demonstrate a crucial role for β-adrenergic receptor signalling by norepinephrine for control of lymphocyte egress from lymph nodes and modulation of cytokine production and proliferation in memory CD8+ memory T cells.181,182
These parallel lines of research have now been joined in studies that demonstrate marked alterations in immune cell activity during cold stress. Nguyen et al. discovered that cold stress stimulates IL-4 and IL-13–driven differentiation of macrophages in brown fat toward an ‘alternative activation’ programme leading to their production of norepinephrine.183 (FIG. 5a). Surprisingly, data obtained using various knock-out mice (deficient in IL-4, IL-13, STAT6 or IL-4 receptor) revealed that the norepinephrine produced by these macrophages is critical for maintaining sufficient thermogenesis in the face of cold stress.183,184 Kokolus et al. further demonstrated that DCs exhibit a reduced ability to stimulate T cells if they are from cold-stressed mice that have a normal body temperature due to increased thermogenesis.183,184 Cold stress is also associated with accelerated tumour growth in murine models, which reflects enhanced tumour cell survival pathways as well as a shifted balance toward an immunosuppressive microenvironment with elevated intratumoral myeloid-derived suppressor cells and regulatory T cells together with reduced CD8+ effector T cells184–186 (FIG. 5b).
(a) Exposure to cold stress drives the release of neurotransmitters, such as norepinephrine, by neurons. This initiates the interleukin-4 (IL-4) and IL-13-driven ‘alternative activation’ programme of differentiation in macrophages, resulting in additional production of norepinephrine, which stimulates β-adrenergic receptors (βAR) expressed on brown adipose cells driving thermogenesis.183 (b) Cold stress in tumour-bearing mice maintained at standard housing temperatures (20–26°C) tilts the balance towards an immunosuppressive local tumour microenvironment. This is characterized by a substantial increase in populations of intratumoral myeloid-derived suppressor cells (MDSCs) and regulatory T (Treg) cells and a concomitant decrease in the number of CD8+ T cells when compared to tumours that develop in mice housed under thermoneutral ambient temperature (30–31°C).184 Tumour cell survival and tumour growth are also accelerated by cold stress. 184–186
An intriguing aspect is that the presence of cancer creates a notable heat seeking behavioural response in animals.184 These data support the conclusions drawn by Romanovsky and colleagues who have contended that endothermic animals, including humans, exhibit heat-seeking behaviour even before other fever-generating symptoms occur.9,10 Findings in this exciting area contribute additional molecular detail to the fundamental role of temperature stress in influencing the functional balance between various arms of the immune system.187,188
Concluding remarks and future directions
The evolutionary conservation of the fever response over millions of years is in line with its protective role — the survival benefit conferred on the host outweighs the metabolic cost of elevating core body temperatures during infection. Cellular components of the immune system have emerged as central players that actively drive fever induction in addition to serving as thermally sensitive effectors. Moreover, the complexity of the molecular pathways that coordinate a febrile response is mirrored by the diverse cell types that are affacted by hyperthermic temperatures; these include DCs, macrophages, NK cells, neutrophils, T and B lymphocytes, and vascular endothelial cells. The picture that emerges is one in which febrile temperatures serve as a systemic alert system that broadly promotes immune surveillance during challenge by invading pathogens. Furthermore, mechanistic insight into the immune-protective nature of fever has opened new avenues to exploit the immunostimulatory activities of thermal stress in the context of cancer therapy.
Fundamental questions remain regarding the nature of the temperature-sensing machinery that triggers changes in immune cell behaviour. HSF1-regulated HSPs are strong candidates in view of their rapid induction even at the relatively modest temperature elevation (ΔT~1–4°C) that accompanies fever.26,122,189–191 Also intriguing are reports that HSF1 regulates additional genes relevant to the induction and/or effector phases of fever, including IL6 and COX2.166,192 Notably, HSP90 and the JAK1–JAK2–STAT3 signalling axis triggered by IL-6 are participants in a feed-forward loop — IL-6-–STAT3 signalling stimulates HSP90 production while JAK2 and STAT3 are established client proteins that are chaperoned by HSP90.193–197 Thus, it is tempting to speculate that induction of HSP90 or other HSPs by febrile temperatures lowers the threshold for IL-6 signalling. Additionally, a class of temperature-sensing transient receptor potential (TRP) cation channel proteins expressed on immune cells and endothelial cells are likely to coordinate responses to febrile temperatures and inflammatory cytokines such as IL-6 and lipid mediators.26,198,199 There are also outstanding questions regarding the mechanisms underlying the spatial regulation by IL-6 during fever induction and lymphocyte trafficking in HEVs. Although brain endothelial cells and HECs are predicted to be main targets for IL-6, contributions of intermediary cells have not been excluded.
Another unresolved question is whether febrile temperatures mobilize innate and adaptive immune cells to sites of infection. Observations that administration of fever-range hyperthermia is effective in boosting E-selectin, P-selectin and ICAM1-dependent trafficking of cytotoxic CD8+ T cells in tumour tissues142 raise the strong possibility that similar mechanisms are enacted by fever in infected tissues. Similarly to CXCL8, several inflammatory chemokines that recruit NK cells, CD4+ and CD8+ T cells, and monocytes (CXCL9, CXCL10, CXCL11, CXCL12) contain putative HSF1-binding sites within their gene promoters and, consequently, may be induced by thermal stress.26 An important caveat is that information regarding the cytokine circuitry (for example, IL-6, RANKL and IL-1) leading to fever as well as the impact of temperature on immune function is largely based on experimental models employing LPS or fever-range hyperthermia as surrogates for pathogen-induced fever. Although these studies provide insight into the mechanistic underpinnings for immune regulation by temperatures within the febrile range, lessons learned from studies of thermogenesis183–185 indicate that overall temperature sensing (cold or hot) in the absence of disease can have unexpected outcomes on innate and adaptive immunity. The next frontier will be to establish whether the same mechanisms identified during challenge of healthy animals with LPS or fever-range hyperthermia are operative during febrile responses to pathogens.
Acknowledgments
We thank M. Appenheimer, J. Black, and M. Messmer for helpful comments on the manuscript, E. Smith and UC Berkeley Natural Resources Library for assistance with archived citations, and J. Muhitch for providing the photomicrograph depicting lymph node HEV. This work was supported by the US National Institutes of Health (CA79765, CA085183, and AI082039) and the Jennifer Linscott Tietgen Family Foundation. We also acknowledge the significant contributions of colleagues in the field that could not always be cited due to space limitations.
Footnotes
Conflict of interest statement: The authors have no conflicting financial interests.






