Neural circuits mediating circulating interleukin-1β-evoked fever in the absence of prostaglandin E2 production

Abstract

Infectious diseases and inflammatory conditions recruit the immune system to mount an appropriate acute response that includes the production of cytokines. Cytokines evoke neurally-mediated responses to fight pathogens, such as the recruitment of thermoeffectors, thereby increasing body temperature and leading to fever. Studies suggest that the cytokine interleukin-1β (IL-1β) depends upon cyclooxygenase (COX)-mediated prostaglandin E₂ production for the induction of neural mechanisms to elicit fever. However, COX inhibitors do not eliminate IL-1β-induced fever, thus suggesting that COX-dependent and COX-independent mechanisms are recruited for increasing body temperature after peripheral administration of IL-1β. In the present study, we aimed to build a foundation for the neural circuit(s) controlling COX-independent, inflammatory fever by determining the involvement of brain areas that are critical for controlling the sympathetic outflow to brown adipose tissue (BAT) and the cutaneous vasculature. In anesthetized rats, pretreatment with indomethacin, a non-selective COX inhibitor, did not prevent BAT thermogenesis or cutaneous vasoconstriction (CVC) induced by intravenous IL-1β (2 μg/kg). BAT and cutaneous vasculature sympathetic premotor neurons in the rostral raphe pallidus area (rRPa) are required for IL-1β-evoked BAT thermogenesis and CVC, with or without pretreatment with indomethacin. Additionally, activation of glutamate receptors in the dorsomedial hypothalamus (DMH) is required for COX-independent, IL-1β-induced BAT thermogenesis. Therefore, our data suggests that COX-independent mechanisms elicit activation of neurons within the DMH and rRPa, which is sufficient to trigger and mount inflammatory fever. These data provide a foundation for elucidating the brain circuits responsible for COX-independent, IL-1β-elicited fevers.

1. Introduction

Pathogens or their fragments induce the production of cytokines by immune cells, causing a systemic inflammatory condition. Interleukin-1β (IL-1β) is a cytokine that triggers neuroimmune-mediated acute-phase responses (Dinarello and van der Meer 2013, Libby 2021) that help the body to fight infection. IL-1β induces fever by increasing heat production in brown adipose tissue (BAT) (Dascombe et al., 1989) and potentially in the heart via tachycardia, and reducing heat loss via cutaneous vasoconstriction (CVC) (Haefeli et al., 1993, Ohashi and Saigusa 1997), thereby increasing core body temperature (TCORE).

Systemic inflammation induced by lipopolysaccharide (LPS) is the most well studied experimental model for bacteremia-induced fever. LPS is a component of the cell wall of Gram-negative bacteria that activates toll-like receptor 4 (TLR4). The complex responses triggered by systemic administration of LPS include the production of cytokines, chemokines, and eicosanoids by immune cells from the blood and peripheral organs. For LPS-induced fever, the transduction of the immune signal to neuronal activity requires a specific eicosanoid: prostaglandin (PG) E₂. The cyclooxygenase enzyme (COX)-2 catalyzes the conversion of the fatty acid arachidonic acid to PGH₂, which can be converted to PGE₂ (Wang et al., 2021). The initial stage of LPS-induced fever requires PGE₂ produced by macrophages in the liver and lungs (Steiner et al., 2006, Garami et al., 2018) and/or by brain endothelial cells (Blomqvist and Engblom 2018) and the late stages require COX-2 activation and the production of PGE2 in brain endothelial cells (Matsumura et al., 1998, Yamagata et al., 2001) and perivascular cells (Elmquist et al., 1997). In contrast to the many studies elucidating the mechanisms underlying LPS-evoked fever little is known about the initiation and maintenance of fever caused by IL-1. Similar to LPS, IL-1 also induces COX-2 activity in brain endothelial cells (Cao et al., 1996, Konsman et al., 2004).

PGE2 binds to EP3 receptors (EP3R) on neurons within the preoptic area of the hypothalamus (POA), leading to the activation of the sympathetic premotor neurons (SPmN) in the rostral raphe pallidus area (rRPa) (Morrison and Nakamura 2019, Machado et al., 2020), which control the sympathetic outflows to BAT (Morrison and Nakamura 2019, da Conceicao et al., 2020, Machado et al., 2020) and the cutaneous vasculature (Ootsuka and Tanaka 2015, McAllen and McKinley 2018). The dorsomedial hypothalamus (DMH), which includes the dorsal aspect of the dorsomedial hypothalamic nucleus and the dorsal hypothalamic area, is a critical sympathoexcitatory relay in the complex neurocircuitry modulating the sympathetic outflow to BAT and to the heart during thermal challenges, COX-2/PGE2-mediated fever, and stress-induced hyperthermia (Kataoka et al., 2014, Machado et al., 2018, Madden and Morrison 2019, Nakamura and Morrison 2022). POA neurons contribute to the COX-2/PGE2-mediated activation (or disinhibition) of neurons residing in the DMH (da Conceicao et al., 2020) and rRPa (Machado et al., 2020). Activity of excitatory amino acid receptors within the DMH is required for BAT thermogenesis but not for cutaneous vasoconstriction, while activation of those receptors within the rRPa is required for both BAT thermogenesis and cutaneous vasoconstriction in response to administration of PGE2 in the POA (Madden and Morrison 2003, Madden and Morrison 2004, Rathner et al., 2008, Nakamura et al. 2009, McAllen and McKinley 2018).

In contrast to the well described neural pathway(s) responsible for COX-2/PGE₂-dependent fever, the pathway(s) mediating COX-independent components of fever are not known. Elucidation of the COX-independent mechanisms may lead to novel therapeutic alternatives, such as monoclonal antibodies and IL-1 receptor antagonists, or other critical downstream mediators that have yet to be determined, to treat conditions in which COX inhibitors are ineffective or contra-indicated, including for some viral infections, such as coronavirus disease 2019 (COVID-19) (Baz and Boivin 2019, FitzGerald 2020, Scavone et al., 2020) and dengue fever (Kellstein and Fernandes 2019). In addition, these therapeutic alternatives will likely avoid side effects related to the use of COX inhibitors, such as cardiovascular toxicity, hepatotoxicity, exacerbated hypersensitivity reactions, and hypotension, especially in critically ill patients, and potentially deadly conditions, as in aspirin-exacerbated respiratory disease Moore et al., 2007, Knights et al., 2010, Dona et al., 2020) Moore et al., 2007, Knights et al., 2010, Velazquez and Teran 2013, Dona et al., 2020). Moreover, COX-2 inhibition in critically ill patients may have detrimental effects, such as increasing the mortality rate under certain conditions (Boyle et al., 1997, Schulman et al., 2005, Watkins et al., 2006, Hui et al., 2010). These undesirable effects of COX inhibition may result from the absence of an important negative feedback regulatory mechanism associated with a PGE2-induced anti-inflammatory effect (Nemeth et al., 2009, Prockop 2013, Simm et al., 2016).

Recent studies and clinical trials using IL-1β monoclonal antibodies showed promising results for the treatment of Mediterranean fever and inflammatory conditions (Dinarello et al., 2012, Dinarello and van der Meer 2013) and are currently being tested for the treatment of severe forms of COVID-19 (current clinical trials in the United States: NCT04362813, NCT04365153, NCT04510493). In addition, recurrent fevers are one of the symptoms in autoinflammatory conditions, which are characterized by dysregulation of the innate immune system largely mediated by IL-1 cytokine family and thus may benefit from clinical inhibitors of IL-1β or its downstream mediators. The potential for the clinical use of inhibitors of IL-1β underscores the need for preclinical studies to identify the effects of IL-1β on critical functions such as autonomic regulation and metabolism.

Pretreatment with a COX inhibitor does not abolish IL-1β-induced fever in rats (De Souza, Cardoso et al., 2002), mice (Sanchez-Alavezet al., 2006), and rabbits (Hashimoto et al., 1988, Murakami et al., 1990, Ohashi and Saigusa 1997). Similarly, genetic deletion of COX-2 in brain endothelial cells attenuates but does not completely abrogate IL-1β-induced fever (Wilhelms et al., 2014) or LPS-induced fever (Nilsson et al., 2017). In addition, EP3R-deficient mice develop the early but not the late phase of fever in response to IL-1β (Sanchez-Alavez et al., 2006). These data suggest that COX-2/PGE2 is not the sole mechanism mediating IL-1β-induced fever. Interestingly, previous studies revealed that several mediators induce increases in body temperature that are resistant or partially resistant to COX inhibitors, such as corticotropin-releasing hormone, IL-8, macrophage inflammatory protein-1, morphine, preformed pyrogenic factor, and endothelin-1 (Minano et al., 1991, Zampronio et al., 1994, Zampronio et al., 2000, De Souza et al., 2002, Fabricio et al., 2005a, Fraga et al., 2008).

In the present study, we sought to determine the brain areas controlling the sympathetic outflow to BAT and the cutaneous vasculature that contribute to COX-independent, IL-1β-induced fever.

2. Materials and methods

2.1. Animals

Male (n = 46) and female (n = 25) Sprague-Dawley rats (290–510 g; Charles River Laboratories, Wilmington, MA, USA) were submitted to the experimental protocols of the present study. Male and female rats display similar IL-1β-induced fever (Mouihate et al., 1998), thus rats from both sexes were randomly distributed across all experimental groups. The animals were kept in a room at 22 °C and under a 12:12 h light/dark cycle with food (LabDiet, St. Louis, MO, USA) and water provided ad libitum. All animal procedures were performed following the regulations in the Guide for the Care and Use of Laboratory Animals (8th ed.; National Research Council, National Academies Press, 2011) and approved by the Animal Care and Use Committee of the Oregon Health and Science University (protocol: TR01_IP00000633).

2.2. Drugs

IL-1β (2 μg/ml; recombinant rat IL-1β/IL-1F2 protein; R&D Systems, Minneapolis, MN, USA) and muscimol (1.2 mM; Tocris, Bristol, UK) were diluted in sterile phosphate buffered saline (PBS). Indomethacin (5 mg/ml; Sigma-aldrich, San Luis, MO, USA), a non-selective COX inhibitor, was diluted in a solution of Tris-HCl (1:10, v/v), pH 8.2, and sterile PBS. The ionotropic glutamate receptor antagonists (iGluRA) (2R)-amino-5-phosphonovalericacid (AP5; 5 mM) and 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX; 5 mM) were mixed in the same solution using sterile PBS. The doses of muscimol and the ionotropic glutamate receptor antagonists were based on previous publications demonstrating inhibitions of thermogenic stimuli, including PGE2-evoked BAT thermogenesis (Madden and Morrison 2003, Madden and Morrison 2004, da Conceicao et al., 2020). The dose of IL-1β used in this study elicits fever (Hansen et al., 2001). The dose of indomethacin prevents LPS-induced fever (Sugimoto et al., 1999, Steiner and Branco 2000, Sprague et al., 2004, Bexis and Docherty 2009, Lkhagvasuren et al., 2011) and, even at a lower dose (2 mg/kg), prevents increases in PGE₂ levels in the CSF in response to zymosan and endothelin-1 (Fabricio et al., 2005b, Bastos-Pereira et al., 2017).

2.3. BAT SNA recordings

Rats were anesthetized by inhalation of 2%-3% isoflurane in 100% O₂ and maintained under isoflurane anesthesia while the femoral vein and artery were cannulated for drug injection and monitoring of arterial pressure (AP) and heart rate (HR), respectively. Rats were then transitioned to intravenous (iv) anesthesia with urethane and α-chloralose (750 and 60 mg/kg, respectively). The trachea was cannulated and connected to a mechanical ventilatory system (~1 ml/100 g body weight, 59–65 S/min, 100% O₂) coupled to a capnograph (Novametrix, Wakefield, MA, USA). Rats were positioned in a stereotaxic apparatus coupled to a spinal clamp to avoid movement-related artifacts in the nerve recordings. The skin was maintained at a warm temperature (~36–38 °C) by a water-perfused blanket and a heat lamp, and skin temperature (TSKIN) was monitored continuously by using a thermocouple placed between the skin and the blanket. Thermocouples were inserted in the rectum to assess TCORE and in the BAT pad to assess BAT temperature (TBAT). A sympathetic nerve innervating the interscapular BAT pad was recorded using bipolar hook electrodes. To avoid reflex movements, the animals received the paralyzing D-tubocurarine (0.6 mg/rat, iv, supplemented every 1–3 h with 0.3 mg). BAT SNA was amplified 5,000–50,000 times (CyberAmp 380, Axon instruments, San Jose, CA, USA), filtered (1–300 Hz), digitized and recorded using Spike2 software (Cambridge Electronic Design, Cambridge, UK). The skin over the skull was opened and holes were drilled in the skull to access target areas. The incisor bar was adjusted at the horizontal level for each animal to obtain a flat skull surface. Nanoinjections (60–100 nl injected within approximately 20 s) using glass micropipettes (20–40 μm tip diameter) connected to the stereotaxic frame, a pressure system (Picospritzer II, General valve corporation, Fairfield, NJ, USA), and a surgical microscope, were performed by positioning the micropipette at 2.8 to 3.5 mm caudal to lambda and 9.2 to 9.7 ventral to dura mater, on the midline, to target the rRPa, or 3.0 to 3.5 mm caudal to bregma, 7.8 to 8.2 ventral to dura mater and 0.3 to 0.4 mm lateral to the midline, to bilaterally target the DMH. The targeted sites were marked using fluorescent beads (1:100, v/v, Molecular Probes, Eugene, OR, USA). After the experiment, the animals were euthanized and the brains were collected and post-fixed in 5% paraformaldehyde prior to sectioning on a microtome to obtain 60 μm-histological sections that were mounted on slides and photographed.

2.4. Forepaw cutaneous temperature recordings

The anesthetic drug urethane induces potent vasoconstriction (Malkinson et al., 1988, Malkinson et al., 1993). Therefore, in order to investigate the vasoconstrictor effects of IL-1β, the animals were anesthetized by inhalation (2%–3% isoflurane in 100% O₂) and maintained under anesthesia after brain surgery (1%–1.25% isoflurane in 100% O₂, via a nose cone) throughout the experiments. Anesthetized, free-breathing rats had their femoral vein and artery cannulated and were positioned in a stereotaxic apparatus. The skin was maintained at a warm temperature (~37.5–38.5 °C) by a water-perfused blanket and a heat lamp. TSKIN and TCORE were monitored as described above. Paw temperature (TPAW) was recorded using a thermocouple taped to the skin surface of one of the forepaws. Brain surgery and nanoinjections were performed as described above.

2.5. Immunoassay for determining plasma PGE₂ levels

The effectiveness of our in vivo indomethacin treatment for preventing the production of PGE₂ was assessed by measuring PGE₂ in arterial blood collected in heparin coated tubes containing indomethacin (10 μM final concentration) and centrifuged (2500g for 10 min at 4 °C). The supernatant was collected and maintained at −70 °C until processing. The plasma samples were used in an enzyme-linked immunosorbent assay (ELISA) for detection and quantification of PGE₂ using a commercial kit (#514010, Cayman chemical, Ann Arbor, MI, USA) that was performed according to the instructions provided by the manufacturer. Prior to pipetting plasma samples on the microplate, plasma samples were diluted 2 or 4-fold in ELISA buffer provided by the manufacturer, therefore, a dilution factor of 2 or 4, respectively, was used to correct the final concentrations of PGE₂. Samples were run in duplicate and the detection limits of the assay ranged from 7.8 to 1,000 pg/ml. Plasma samples from three rats deviated from confidence analysis criteria by the ELISA kit manufacturer and were excluded from the study.

2.6. Experimental design

The animals were instrumented for recording of either BAT SNA or forepaw cutaneous temperature. In both experimental conditions, animals received IL-1β (2 μg/kg, 2 μg/ml, iv). Rats received saline (1 ml/kg, iv) or indomethacin (5 mg/kg, 5 mg/ml, iv) 30–45 min prior to IL-1β. To inhibit the activity of neurons in rRPa, the GABA_A receptor agonist, muscimol (1.2 mM, 60–80 nl), was nanoinjected in the rRPa shortly after observing the onset of IL-1β-evoked increases in BAT SNA, which occurred consistently ~ 10–20 min post-IL-1β. For cutaneous temperature recordings, muscimol was nanoinjected in the rRPa after IL-1β or prior to IL-1β in the animals pretreated with indomethacin. These experiments were designed to investigate the effect of the inhibition of sympathetic premotor neurons residing in the rRPa in preventing and reversing IL-1β-evoked vasoconstriction.

To evaluate the contribution of iGluR in the DMH for indomethacin-resistant IL-1β-evoked BAT thermogenesis, a group of rats received indomethacin 30–45 min prior to IL-1β and, after observing increases in BAT SNA and BAT temperature of ≥ 1.5 °C, a solution containing AP5 and CNQX (5 mM for each compound) was nanoinjected bilaterally (100 nl in each side) in the DMH. At the end of the experiments, from 120 to 140 min after administration of IL-1β, arterial blood samples were collected for further analysis of PGE₂ levels.

A separate group of awake rats received an intraperitoneal (ip) injection of saline (1 ml/kg, ip) or indomethacin (5 mg/kg, ip) followed 30 min later by IL-1β (2 μg/kg, ip). Two hours after administration of IL-1β, rats were briefly (for ~ 15 min) anesthetized with urethane and α-chloralose (750 and 60 mg/kg, respectively, ip) prior to blood collection for analysis of PGE₂ levels) via a femoral artery and subsequent euthanasia.

2.7. Data and statistical analyses

The amplitude of BAT SNA was calculated every 4 s as the root mean square value of the total power in the 0.1 to 20 Hz band of the auto-spectrum of the raw BAT SNA signal. For each rat, the baseline (BL) level of BAT SNA was determined as the mean BAT SNA amplitude during a 30 s period when the rat was warm and the BAT SNA was at a minimum level. The BAT SNA values are presented as % BL where mean BAT SNA values over 30 s periods are normalized to the 30 s BL values, as we described previously (Mota et al., 2020). Pre-IL-1β and peak values of the variables were obtained during the 30 s preceding and during the 30 s of the maximum responses of the variables within 40 min after administration of IL-1β, respectively. For the groups that received muscimol in the rRPa, BAT SNA peak values are the maximum mean values over 30 s within 5 min after nanoinjection of muscimol. For CVC experiments, nadir and peak values of TPAW are the minimum and maximum mean values, respectively, over 30 s within 60 min after administration of IL-1β. For the group that received iv IL-1β and muscimol in the rRPa, we analyzed three repeated measures: 30 s mean values: 1) before IL-1β, 2) after IL-1β and immediately preceding the nanoinjection of muscimol, and 3) peak values within 30 min after the nanoinjection of muscimol. For the groups that received AP5 and CNQX in the DMH following IL-1β, we analyzed three repeated measures: 30 s mean values: 1) before IL-1β, 2) after IL-1β and immediately preceding the nanoinjections of AP5 and CNQX, and 3) nadir values within 10 min after the nanoinjections. The area under the curve of IL-1β-evoked increases in BAT SNA from 0 to 40 min after administration of IL-1β was calculated with the baseline value defined as zero.

All statistical analyses were performed using SPSS software platform (IBM; Armonk, NY, USA). Data from two repeated measures within the same group were analyzed using paired Student’s t test and data from two measures in different groups were analyzed using independent-samples Student’s t test in accordance with Levene’s test for equality of variances. Data from three repeated measures were analyzed using repeated measures one-way ANOVA followed by Bonferroni or Fisher’s Least Significant Difference tests in accordance with Mauchly’s test of sphericity and Greenhouse-Geissner correction when appropriate. Data from more than two groups were analyzed using one-way ANOVA followed by Bonferroni post hoc test in accordance with Levene-Brown-Forsythe’s test for equality of variances. Results are expressed as mean ± standard deviation (SD) and statistical results of p < 0.05 were considered significant.

3. Results

3.1. IL-1β-evoked BAT thermogenesis, tachycardia and hypertension are resistant to the pharmacological inhibition of COX by indomethacin

Initially, we sought to determine the effect of a single iv injection of IL-1β on BAT thermogenesis. Administration of IL-1β following iv saline pretreatment (Fig. 1A and 1B) increased BAT SNA (+2852 ± 2020% BL, t₍₈₎ = 4.23, p = 0.003), TBAT (+2.6 ± 1.3 °C, t₍₈₎ = 6.05, p < 0.001), expired CO₂ (+1.0 ± 0.4%, t₍₈₎ = 7.17, p < 0.001), TCORE (+0.6 ± 0.2 °C, t₍₈₎ = 7.85, p < 0.001), HR (+102 ± 42 bpm, t₍₈₎ = 7.3, p < 0.001) and mean AP (MAP) (+11 ± 7 mmHg, t₍₈₎ = 4.44, p = 0.002). Our next step was to determine if IL-1β-was still capable of evoking responses following pretreatment with indomethacin. Following pretreatment with indomethacin (Fig. 1C and D), iv IL-1β evoked increases in BAT SNA (+2545 ± 756% BL, t₍₆₎ = 8.91, p < 0.001), TBAT (+2.9 ± 1.3 °C, t₍₆₎ = 6.17, p < 0.001), expired CO₂ (+1.2 ± 0.6%, t₍₆₎ = 4.85, p = 0.003), TCORE (+0.9 ± 0.5 °C, t₍₆₎ = 4.37, p = 0.005), HR (+117 ± 47 bpm, t₍₆₎ = −6.6, p < 0.001) and MAP (+27 ± 20 mmHg, t₍₆₎ = 3.68, p = 0.01).

Fig. 1.

IL-1β-evoked BAT thermogenesis is not inhibited by indomethacin, a non-selective COX inhibitor. A, C, Representative examples of an intravenous injection of IL-1β increasing BAT SNA, TBAT, expired CO₂, TCORE, HR and AP in saline-pretreated (A) and in indomethacin-pretreated (C) rats. Saline or indomethacin was given 30–45 min prior to the IL-1β injection. Red traces represent the raw BAT SNA signal recordings. B, D, Expanded BAT SNA recordings showing the action potential compounds from saline-pretreated (a, b) and in indomethacin-pretreated (c, d) rats, before (a, c) and at the peak (b, d) of IL-1β-evoked increases in BAT SNA shown in C and A, respectively. Indo., indomethacin. Data represented as mean ± SD are shown in Fig. 3.

The latency to the onset of IL-1β-evoked increases in BAT SNA (defined as the time at which the increase in BAT SNA exceeded BL by 2-fold) did not differ between the groups that were pretreated with saline (7.6 ± 6.4 min) or indomethacin (12.0 ± 7.0 min, p = 0.16, t(_17.76) = 1.47). Moreover, the area under the curve of the IL-1β-evoked increases in BAT SNA was not different between the groups pretreated with saline (94,734.6 ± 77,363.5 %BL*s) or indomethacin (101,732.5 ± 44,572.7 %BL*s, p = 0.85, t(₁₂) = 0.1972).

3.2. Nanoinjection of muscimol, a GABA_A receptor agonist, in the rRPa inhibits IL-1β-evoked BAT thermogenesis in saline-pretreated and in indomethacin-pretreated rats

We next investigated if IL-1β requires activation of BAT SPmN in rRPa, which would suggest that IL-1β recruits a neural circuit preceding activation of those neurons. There were no significant differences among the groups prior to IL-1β injection (Table 1), with the exception of MAP. Nonetheless the values for MAP in all groups were comparable with baseline MAP values for Sprague Dawley rats observed in other studies (Schreihofer et al., 2005, Wang et al., 2013).

Table 1.

Pre-interleukin (IL)-1β administration values of BAT SNA shown as % baseline (% BL), TBAT, expired CO₂, TCORE, HR and MAP.

Variable	IL-1β (n = 9)	Indomethacin + IL-1β (n = 7)	IL-1β + muscimol (rRPa) (n = 5)	Indomethacin + IL-1β + muscimol (rRPa) (n = 7)	One-way ANOVA	p
BAT SNA (% BL)	128 ± 51	177 ± 164	112 ± 16	108 ± 15	F(3,27) = 0.86	0.47
TBAT (°C)	33.3 ± 1.4	34.3 ± 1.0	33.9 ± 1.0	34.5 ± 0.7	F(3,27) = 1.68	0.20
Expired CO2 (%)	3.9 ± 0.4	3.6 ± 0.5	3.9 ± 0.2	3.8 ± 0.3	F(3,27) = 2.90	0.06
TCORE (°C)	36.6 ± 0.5	37.0 ± 0.4	36.8 ± 0.6	37.2 ± 0.4	F(3,27) = 2.01	0.14
HR (bpm)	349 ± 33	373 ± 53	358 ± 19	407 ± 61	F(3,27) = 1.12	0.36
MAP (mmHg)	117 ± 12	97 ± 17	89 ± 23*	91 ± 16*	F(3,27) = 4.84	0.009

Results are expressed as mean ± SD. N = 5–9/group.

^*p < 0.05 vs IL-1β, using one-way ANOVA followed by Bonferroni post hoc test.

Within group analyses showed that nanoinjection of muscimol in the rRPa (Fig. 2B) prevented (Fig. 2A) IL-1β-evoked increases in BAT SNA (+8 ± 29% BL vs. pre-IL-1β, t₍₄₎ = 0.81, p = 0.46), TBAT (+0.2 ± 0.1 °C vs. pre-IL-1β, t₍₄₎ = −0.56, p = 0.61), expired CO₂ (+0.2 ± 0.1% vs. pre-IL-1β, t₍₄₎ = 2.83, p = 0.05), TCORE (+0.0 ± 0.0 °C vs. pre-IL-1β, t₍₄₎ = −1.96, p = 0.12), MAP (+15 ± 17 mmHg vs. pre-IL-1β, t₍₄₎ = 1.06, p = 0.35) and HR (+72 ± 50 bpm vs. pre-IL-1β, t₍₄₎ = 1.45, p = 0.22).

Fig. 2.

Nanoinjection of muscimol, a GABA_A receptor agonist, in the rRPa inhibits IL-1β-evoked BAT thermogenesis in the presence of indomethacin. A, C, Representative examples of a nanoinjection of muscimol in the rRPa inhibiting IL-1β-evoked increases in BAT SNA, TBAT, expired CO₂ and TCORE in saline-pretreated (A) and in indomethacin-pretreated (C) rats. Saline or indomethacin was given 30–45 min prior to the IL-1β injection. Red traces represent the raw BAT SNA signal recordings. B, D, Photomicrographs of representative examples of muscimol injection sites in the rRPa and diagram drawings depicting the anatomical locations of all the muscimol injection sites in the rRPa saline-pretreated (B) and in indomethacin-pretreated (D) rats at approximately 11–11.7 mm caudal to bregma. Indo., indomethacin. rRPa, rostral raphe pallidus area. 7, facial motor nucleus. Data represented as mean ± SD are shown in Fig. 3.

Similarly, nanoinjection of muscimol in the rRPa (Fig. 2D) in indomethacin-pretreated rats (Fig. 2C) prevented IL-1β-evoked increases in BAT SNA (+76 ± 154% BL vs. pre-IL-1β, t₍₆₎ = 1.39, p = 0.21), TBAT (+0.2 ± 0.2 °C vs. pre-IL-1β, t₍₆₎ = 0.68, p = 0.52), expired CO₂ (+0.1 ± 0.1% vs. pre-IL-1β, t₍₆₎ = 0.12, p = 0.31), TCORE (+0.0 ± 0.0 °C vs. pre-IL-1β, t₍₆₎ = 0.46, p = 0.66), HR (+23 ± 45 bpm vs. pre-IL-1β, t₍₆₎ = 2.10, p = 0.8) and MAP (+14 ± 21 mmHg vs. pre-IL-1β, t₍₆₎ = 1.51, p = 0.18).

Comparisons between groups (Fig. 3) using one-way ANOVA showed significant overall effects for BAT SNA (F_(3,27) = 10.88, p < 0.001), TBAT (F_(3,27) = 16.12, p < 0.001), expired CO₂ (F_(3,27) = 14.07, p < 0.001), TCORE (F_(3,27) = 15.75, p < 0.001) and HR (F_(3,27) = 4.53, p = 0.01), but not for MAP (F_(3,27) = 1.47, p = 0.25). Compared to IL-1β alone (Fig. 1A), nanoinjection of muscimol in the rRPa (Fig. 2A) markedly inhibited (Fig. 3) IL-1β-evoked increases in BAT SNA (p < 0.01), TBAT (p < 0.01), expired CO₂ (p < 0.01) and TCORE (p < 0.01), but not in HR (p = 0.08). Similarly, compared to indomethacin-pretreated rats that received IL-1β (Fig. 1C), nanoinjection of muscimol in the rRPa in indomethacin-pretreated rats (Fig. 2C), markedly inhibited (Fig. 3) IL-1β-evoked increases in BAT SNA (p < 0.01), TBAT (p < 0.01), expired CO₂ (p < 0.01) and TCORE (p < 0.01), but not in HR (p = 0.12). The demonstration that the IL-1β-induced responses require activation of rRPa neurons suggests that iv IL-1β recruits a neural circuit that precedes and drives activation of BAT SPmN in rRPa.

Fig. 3.

Indomethacin-treated groups show similar IL-1β-evoked BAT thermogenesis and cardiovascular responses when compared to their respective control groups. The quantifications of the changes in the measured variables (BAT SNA, TBAT, expired CO₂, TCORE, HR and MAP) were calculated as the maximum values within 40 min after administration of IL-1β minus the pre-IL-1β values. Open and black circles represent saline-pretreated and indomethacin-pretreated rats, respectively. Open and black squares represent saline-pretreated and indomethacin-pretreated rats, respectively, that received a nanoinjection of muscimol in the rRPa after the first compound potentials evoked by IL-1β. % BL, % baseline. Indo., indomethacin. Data are represented as mean ± SD. N = 5–9/group *p < 0.05 vs saline + IL-1β and #p < 0.05 vs indomethacin + IL-1β, using one-way ANOVA followed by Bonferroni post hoc test.

For all of the analyzed variables, the responses evoked by iv IL-1β were not significantly different (Figs. 1 and 3) between the saline-pretreated and indomethacin-pretreated groups (post hoc between-subjects comparisons using Bonferroni tests) (Table 2). These data show that the IL-1β-induced increases in thermogenic variables were not attenuated by COX inhibition with indomethacin.

Table 2.

Effect of systemic administration of IL-1β on BAT and cardiovascular variables.

	Saline + IL-1β (n = 5)	Indomethacin + IL-1β (n = 7)	p value of post hoc test
BAT SNA (% BL)	+2852 ± 2020	+2545 ± 756	1.00
TBAT (°C)	+2.6 ± 1.3	+2.9 ± 1.3	0.50
Exp CO2 (%)	+1.0 ± 0.4	+1.2 ± 0.6	1.00
TCORE (°C)	+0.6 ± 0.2	+0.9 ± 0.5	1.00
HR (bpm)	+102 ± 42	+117 ± 47	1.00
AP (mmHg)	+11 ± 7	+27 ± 20	0.35

Results are expressed as mean ± SD. One-way ANOVA followed by Bonferroni post hoc test included the four groups analyzed for BAT SNA (see Fig. 3).

3.3. IL-1β-evoked cutaneous vasoconstriction is resistant to the pharmacological inhibition of COX by indomethacin and requires activation of rRPa neurons

Fever is not only driven by increases in BAT thermogenesis but also by increases in CVC, thereby reducing heat loss to the environment. In our experiments, we monitored TPAW as an index of the sympathetic tone to the skin vasculature, with elevated TPAW representing a vaso-dilated state and decreases in TPAW representing increased CVC tone.

There were no significant differences in TPAW or TCORE among the groups prior to IL-1β injection (Table 3) or in the latency to the onset of CVC following IL-1β injection, which was defined as a decrease of 1 °C in TPAW (12.7 ± 10.3 min for saline + IL-1β, vs 9.3 ± 5.9 min for indomethacin + IL-1β, p = 0.49, t(_9.68) = 0.72). Repeated measures one-way ANOVA showed a significant overall effect for TPAW (F_(2,14) = 106.99, p < 0.001), but not for TCORE (F_(2,14) = 3.438, p = 0.07). Bonferroni tests of within-subjects comparisons showed that administration of IL-1β (Fig. 4A, B and D) markedly reduced TPAW (−6.9 ± 1.2 °C post-IL-1β vs pre-IL-1β, p < 0.001). Nanoinjection of muscimol in the rRPa (Fig. 4C) during paw vasoconstriction (Fig. 4B) reversed the IL-1β-evoked CVC to pre-IL-1β levels (Fig. 4D; TPAW: +6.9 ± 1.4 °C, post-IL-1β vs post muscimol, p < 0.001).

Table 3.

Pre-interleukin (IL)-1β administration values of cutaneous forepaw temperature (TPAW) and TCORE.

Variable	IL-1β (n = 5)	Indomethacin + IL-1β (n = 5)	Indomethacin + IL-1β + muscimol (rRPa) (n = 3)	One-way ANOVA	p
TPAW (°C)	35.5 ± 0.5	34.6 ± 1.8	35.3 ± 0.5	F(2,12) = 0.71	0.51
TCORE (°C)	37.7 ± 0.7	37.3 ± 0.6	37.7 ± 0.2	F(2,12) = 0.86	0.45

Results are expressed as mean ± SD. N = 3–5/group. Comparisons between groups were performed by using one-way ANOVA.

Fig. 4.

IL-1β-evoked vasoconstriction is not prevented by a COX inhibitor and requires activation of rRPa neurons. A, D, Representative examples of an intravenous injection of IL-1β reducing TPAW in saline-pretreated (A) and in indomethacin-pretreated (E) rats. B, F, Nanoinjection of muscimol in the rRPa reverses IL-1β-evoked vasoconstriction in saline-pretreated rats (B) and prevents IL-1β-evoked vasoconstriction in indomethacin-pretreated (F) rats. Saline or indomethacin was given 30–45 min prior to the IL-1β injection. C, G, Photomicrographs of representative examples of muscimol injection sites in the rRPa and diagram drawings depicting the anatomical locations of all the muscimol injection sites in the rRPa saline-pretreated (C) and in indomethacin-pretreated (G) rats at approximately 11.5 mm caudal to bregma. 7, facial motor nucleus. Indo., indomethacin. rRPa, rostral raphe pallidus area. D, TPAW values before and after the injection of IL-1β showing the induction of CVC and the reversal of CVC after a nanoinjection of muscimol in the rRPa. *p < 0.001 vs pre-IL-1β and #p < 0.001 vs post-muscimol, using one-way ANOVA followed by Bonferroni post hoc test. There was no significant change in TCORE values. H, Nanoinjection of muscimol in the rRPa prevented IL-1β-induced CVC in indomethacin-pretreated rats. The quantifications of the changes in TPAW and TCORE were calculated as the minimum or maximum values, respectively, within 60 min after administration of IL-1β minus the pre-IL-1β values. Circles represent indomethacin-pretreated rats that received iv IL-1β. Squares represent indomethacin-pretreated rats that received a nanoinjection of muscimol in the rRPa prior to iv IL-1β. *p < 0.05 vs indomethacin + IL-1β, independent Student’s t test. Data are represented as mean ± SD. N = 3–5/group.

Comparisons between pre-IL-1β and post-IL-1β values showed that IL-1β also induced CVC in indomethacin-pretreated rats (Fig. 4E), as observed by a reduction in TPAW (−5.9 ± 3.4 °C, t₍₄₎ = 3.87, p = 0.02) and an increase in TCORE (+0.7 ± 0.3 °C, t₍₄₎ = −4.52, p = 0.01). Nanoinjection of muscimol in the rRPa (Fig. 4G) prior to iv IL-1β (Fig. 4F) markedly reduced CVC (TPAW: −1.0 ± 0.3 °C) and febrile (TCORE: +0.0 ± 0.1 °C) responses in comparison to the group that did not receive muscimol in the rRPa (independent t tests: TPAW: t_(4.13) = −3.18, p = 0.03; TCORE: t_(4.98) = 4.08, p = 0.01), which corresponds to an inhibition of approximately 83% of the full CVC response (Fig. 4F and 4H).

3.4. IL-1β-evoked BAT thermogenesis in indomethacin-pretreated rats requires activation of iGluR in the DMH

Repeated measures one-way ANOVA showed significant overall effects for BAT SNA (F_(1,4) = 11.44, p = 0.027), TBAT (F_(2,8) = 154.98, p < 0.001), expired CO₂ (F_(2,8) = 52.67, p < 0.001), HR (F_(2,8) = 44.2, p < 0.001) and MAP (F_(2,8) = 6.27, p = 0.02), but not for TCORE (F_(1,4.1) = 5.2, p = 0.08). Fisher’s Least Significant Difference tests of within-subjects comparisons showed that administration of IL-1β (Fig. 5A and C) increased BAT SNA (+1028 ± 685% BL, p = 0.03), TBAT (+2.3 ± 0.1 °C, p < 0.001), expired CO₂ (+0.7 ± 0.2%, p < 0.001) and HR (+77 ± 18 bpm, p < 0.001), but not MAP (+12 ± 1 mmHg, p = 0.05). Bilateral nanoinjections of AP5 and CNQX (iGluRA) in the DMH (Fig. 5B) markedly reduced IL-1β-evoked increases in BAT SNA (−838 ± 542% BL, p = 0.03), TBAT (−1.1 ± 0.4 °C, p = 0.003), expired CO₂ (−0.4 ± 0.1%, p = 0.009), and HR (−38 ± 20 bpm, p = 0.01) and reduced MAP (−4 ± −2 mmHg, p = 0.02; Fig. 5C). These data suggest that a glutamatergic input to the DMH is necessary for providing the excitatory drive for BAT thermogenesis and tachycardia in response to circulating IL-1β.

Fig. 5.

Blockade of iGluR using AP5 and CNQX in the dorsomedial hypothalamus (DMH) reduces IL-1β-induced BAT thermogenesis and tachycardia in indomethacin-pretreated rats. A, Representative example of the effects of bilateral (R, right, and L, left) nanoinjections of iGluRA (AP5 and CNQX) in the DMH on IL-1β-evoked responses. Red traces represent the raw BAT SNA recordings. B, Photomicrograph of representative example of iGluRA injection sites in the DMH and a diagram drawing depicting the anatomical locations of all the injection sites at approximately 3–3.5 mm caudal to bregma. C, iGluRA in the DMH reduced iv IL-1β-evoked increases in BAT SNA, TBAT, expired CO₂, HR and MAP in indomethacin-pretreated rats. Indomethacin was given 30–45 min prior to the IL-1β injection. DMH, dorsomedial hypothalamus. iGluRA, ionotropic glutamate receptor-antagonists. Indo., indomethacin. Data are represented as mean ± SD. N = 5. *p < 0.05 vs pre-IL-1β, #p < 0.05 vs post-IL-1β.

3.5. IL-1βincreased plasma PGE₂ levels in a COX-dependent manner in awake, but not in anesthetized rats

We assessed the effect of our anesthetic protocols and indomethacin pretreatment on COX activity by determining the plasma levels of PGE₂. Comparisons between groups (Fig. 6) using one-way ANOVA followed by Bonferroni post hoc test showed a significant overall effect of indomethacin and the anesthetics on plasma PGE₂ levels (F_(6,34) = 7.6, p < 0.001). In rats that were not anesthetized when receiving the treatments, systemic administration of IL-1β (2 μg/kg, ip) increased plasma PGE₂ levels compared to the control group, which received indomethacin (5 mg/kg, ip) followed by saline (1 ml/kg, ip), (569.9 ± 384.2 pg/ml vs 118.0 ± 19.4 pg/ml, respectively, p < 0.001). Indomethacin prevented the IL-1β-evoked increases in plasma PGE₂ levels compared to rats that received saline + IL-1β (124.7 ± 26.7 pg/ml, vs. 569 ± 384.2 pg/ml, respectively, p < 0.001). Following indomethacin the plasma PGE₂ levels did not differ between IL-1β and saline treated rats (p = 1.0, Fig. 6).

Fig. 6.

IL-1β increased plasma PGE₂ levels in awake, but not in anesthetized rats, and those increases were prevented by indomethacin. In awake rats, systemic administration of IL-1β increased plasma PGE₂ levels, while indomethacin prevented IL-1β-evoked increases in plasma PGE₂ levels. Plasma PGE₂ levels were lower in saline- and indomethacin-pretreated groups that were maintained under anesthesia during and foll–owing the IL-1β treatment in comparison to a positive control group of awake rats that received saline + IL-1β (*p < 0.01 vs all the other groups, using one-way ANOVA followed by Bonferroni post hoc tests). Indo., indomethacin. Uret./chlo., urethane and α-chloralose anesthesia. Isof., isoflurane anesthesia. Data are represented as mean ± SD. N = 5–9/group.

There were no significant differences in circulating levels of PGE₂ between the groups treated with the two different anesthetic protocols, nor were there significant differences between the saline- and indomethacin-pretreated groups that were maintained under anesthesia (p = 1.0 for each pairwise comparison, Bonferroni tests; Fig. 6), or in comparison to the control group (treated with indomethacin + saline; p = 1.0). In summary, IL-1β did not increase plasma PGE₂ levels in anesthetized rats. In addition, plasma PGE₂ levels following IL-1β were significantly higher (by ~ 5 fold) in rats that were not anesthetized when receiving the IL-1β compared to all other groups of anesthetized rats (Bonferroni post-hoc tests: urethane and α-chloralose anesthetized, saline pretreated: 96.7 ± 30.8 pg/ml, p < 0.001; urethane and α-chloralose anesthetized, indomethacin pretreated: 119.7 ± 51.9 pg/ml, p < 0.001; isoflurane anesthetized, saline pretreated: 82.5 ± 21.2 pg/ml, p < 0.001; isoflurane anesthetized, indomethacin-pretreated: 108.4 ± 39.1 pg/ml, p < 0.001, Fig. 6).

4. Discussion

IL-1β administered systemically elicits a fever by increasing the sympathetic outflow to BAT, thereby stimulating BAT metabolism (monitored as increases in expired CO₂) and BAT thermogenesis. A tachycardia also contributes to the heat production following iv IL-1β administration. Increased heat retention resulting from a sympathetically-mediated increase in CVC complements these thermogenic mechanisms to result in an iv IL-1β-evoked elevation in TCORE that is unaffected by indomethacin pretreatment and thus independent of the products of the enzymatic action of COX. The fever-sustaining increases in the sympathetic outflows to BAT, the heart and the cutaneous vasculature driven by circulating IL-1β require the activity of neurons in the rRPa, likely the populations of BAT, CVC and cardiac SPmN that reside there (Morrison and Nakamura 2019). A glutamatergic activation of the thermogenesis-promoting neurons in the DMH is required for the IL-1β-evoked increases in BAT metabolism and thermogenesis and in heart rate, likely through the excitatory projections from the DMH to the BAT and cardiac SPmN in the rRPa. These results provide not only initial insights into the brain circuits responsible for iv IL-1β-dependent fever, but also a foundation for identifying the brain site(s) at which circulating IL-1β interacts with the IL-1 receptor to initiate the changes in neuronal discharge that produce these fever-supporting autonomic responses.

IL-1β-induced fever is reduced but not eliminated by the COX enzyme inhibitor, indomethacin (Hashimoto et al., 1988, De Souza et al., 2002). The lack of effect of pretreatment with indomethacin on the iv IL-1β-evoked increases in BAT thermogenesis and CVC indicates that these thermoeffector mechanisms contribute to a COX-independent IL-1β-induced fever. In addition, IL-1β-induced responses were similar between male and randomly-cycling female rats, in agreement with a previous study (Mouihate et al., 1998). Nevertheless, further detailed investigation during different phases of the estrous cycle is encouraged and would clarify whether ovarian hormones affect COX-independent fever. Indomethacin effectively blocks endotoxin-induced synthesis of PGE2 and other PGs for at least 5 h (Salvemini et al., 1995), with a half-life of approximately 8.5 h in rats when delivered iv at doses starting at 2.8 mg/kg (Suzuki et al., 1997). Our determinations of the plasma levels of PGE2 in isoflurane or urethane anesthetized and indomethacin-pretreated rats are consistent with previous observations that these anesthetics reduce COX-2 expression and/or PGE2 production in vivo (Martinez et al., 2000, Wang et al., 2014) and in vitro (Wang et al., 2014, Ren et al., 2016). Indeed, our anesthetized rats displayed plasma PGE2 levels comparable to control rats (Park et al., 2015), while plasma PGE₂ levels increased to values comparable to those induced by an intracerebroventricular injection of IL-1β, in awake rats (Park et al., 2015). Together, these data suggest that in our anesthetized rats, COX activity is suppressed to such a degree that administration of the COX inhibitor, indomethacin, did not further suppress plasma PGE₂ levels. Thus, consistent with an earlier demonstration that urethane does not inhibit IL-1 fever (Malkinson et al., 1988), our finding that iv IL-1β induces a robust activation of fever-supporting sympathetic thermoeffectors in anesthetized rats validates this model for elucidating the neural mechanisms underlying a COX-2/PGE₂/EP3 receptor-independent component of IL-1β-induced fever.

IL-1β can indirectly influence neuronal activity by inducing endothelial cells in the brain to produce and release cytokines, prostaglandins, and other mediators (Schiltz and Sawchenko, 2002, Konsman et al., 2004). Additionally, cytokines can diffuse from the circulation to circumventricular organs, where they can act to modulate the activity of brain cells (Hashimoto et al., 1991, Broadwell and Sofroniew, 1993, Maness et al., 1998, Quan et al., 1998, Vitkovic et al., 2000), since IL-1 receptors are found within the central nervous system (Ericsson et al., 1995). Alternatively, IL-1 changes blood–brain-barrier (BBB) permeability (Blamire et al., 2000, Skinner et al., 2009) and can be transported from the blood to the brain via the BBB (Banks et al., 1991, Banks et al., 1993, McLay et al., 2000, Argaw et al., 2006, Erickson et al., 2018). Therefore, in our study using indomethacin, IL-1β could be triggering fever by acting directly on cells within the central nervous system, since indomethacin does not prevent the entry of IL-1 into the brain (Banks et al., 1991).

Our finding that systemic administration of indomethacin prevented IL-1β-induced increases in plasma PGE₂ levels shows that, differently from LPS-induced fever (Garami et al., 2018), the initiation of IL-1β-induced fever does not depend on the entrance of circulating PGE₂ into the brain. Since sharp increases in plasma IL-1β levels occur as early as 15 min after an intraperitoneal injection with IL-1β (Hansen et al., 2001) and peripherally-administered indomethacin does not prevent the entrance of IL-1β into the brain in mice (Banks et al., 1991), a hypothesis remaining to be tested is whether the initiation of IL-1β-induced fever requires circulating IL-1β crossing the BBB, despite the small amount of IL-1β crossing the BBB (Coceani et al., 1988, McLay et al., 2000). In this regard, the potential role of areas lacking a tight BBB, the circumventricular organs, are a particularly attractive target that remains to be fully investigated. Additionally, IL-1β acts on endothelial cells to contribute to fever. IL-1β-induced fever requires IL-1R1 in endothelial cells (Ching et al., 2007, Knoll et al., 2017) which may produce intricate, non-disruptive, effects in the BBB that may lead to synthesis of an intermediate mediator that is necessary for IL-1 fever (Varatharaj and Galea 2017). One mechanism for the initiation of IL-1β-induced fever is that circulating IL-1β activates IL-1R1 in endothelial cells, which produce (Cao et al., 1996, Konsman et al., 2004). Presumably, increased COX-2 activity would increase the local production and release of PGE₂, which binds to EP3R within the POA, thus triggering fever. There is evidence to support the role of PGE₂ in IL-1β-induced fever, especially in late stages of fever, whereas there is conflicting data regarding the early stage of IL-1β fever. Icv administration or a low dose of IL-1β administered iv (100 ng/mouse) do not induce fever in EP3R-deficient mice (Ushikubi et al., 1998). In contrast, it has been shown that EP3R-deficient mice or intact mice that had been pretreated with indomethacin (5 mg/kg, ip) develop the early (0–30 min) but not the late (>30 min) stage of fever in response to a micro-injection of IL-1β in the anterior hypothalamus (Sanchez-Alavez et al., 2006). In our study, thermoeffector responses necessary for IL-1β-induced fever were resistant to a systemic injection with indomethacin, thus showing that these responses did not require peripheral PGE₂. Our findings do not refute or contradict previous studies nor do we suggest that IL-1β-induced fever is entirely COX-independent. Instead, our data demonstrate a COX-independent pathway for the initiation of IL-1β-induced fever and define brain regions necessary for the early stage of this fever. In fact, our data complement the well-established COX-2/PGE₂-dependent pathway that sustains inflammatory fever by revealing that the initiation of IL-1β-induced fever does not require circulating PGE₂.

Although we demonstrated in the current studies that indomethacin (5 mg/kg) blocked the peripheral production of PGE₂, whether peripherally-administered indomethacin effectively blocks the production of PGE₂ in the brain is debated. A study in mice showed that indomethacin (10 mg/kg, ip) reduced peripheral but not brain inflammation induced by carrageenan and that the content of indomethacin in the brain was very low (Gamache and Ellis, 1986). In contrast, indomethacin (2 mg/kg, ip) reduced fever in response to icv arachidonic acid (De Souza et al., 2002) and prevented increases in PGE₂ levels in the CSF in response to zymosan and endothelin-1 (Fabricio et al., 2005b, Bastos-Pereira et al., 2017), suggesting that peripherally-administered indomethacin can prevent the production of PGE₂ in the central nervous system. Whether systemically administered indomethacin also prevents the IL-1β-induced production of PGE₂ centrally remains to be determined.

In vivo and in vitro studies suggest that the early stage of fever after direct delivery of IL-1β in the anterior hypothalamus occurs independently of COX-2 activation but requires activation of a fast-signaling cascade via the IL1R1-MyD88-sphingomyelinase-ceramide pathway that, indirectly, inhibits POA neurons by increasing local GABA release (Nalivaeva et al., 2000, Sanchez-Alavez et al., 2006, Tabarean et al., 2006). If these neurons, similarly to EP3R-expressing POA neurons (Machado et al., 2020), contribute to disinhibit rRPa neurons that regulate BAT thermogenesis and CVC, a fever response is initiated. A direct effect of IL-1β in POA neurons is unlikely, since IL1R1 mRNA is not found in POA neurons, but only in cerebral vasculature in the POA (Ericsson et al., 1995). Our findings are complementary to previous studies and suggest that the early stage of IL-1β-induced fever recruits a COX-independent mechanism requiring activation of neurons in the DMH and rRPa. Despite these findings, there is still a lack of detailed in vivo studies determining the mechanisms underpinning the initial stage of fever caused by IL-1β, which could contribute to the understanding of fevers that are resistant to COX inhibitors and reveal targets for the development of drugs to treat those conditions.

Another plausible hypothesis for the persistence of the sympathetic responses observed in our study is a COX-independent activation of the vagus nerve in response to IL-1β. Indeed, afferent vagal nerve activity is increased by IL-1β (Niijima, 1996, Kurosawa et al., 1997, Ek et al., 1998) and nodose ganglion neurons express IL-1 receptor (Ek et al., 1998), suggesting the possibility of direct effects of IL-1β on vagal afferent activity.

Increased heat retention via CVC plays a more significant role in fever development in relatively hairless humans than in small, hairy rodents. The sympathetic control of the blood flow through cutaneous vascular beds is regulated by spinal sympathetic CVC preganglionic neurons that receive their principal excitatory drives from CVC SPmN in the rRPa and in the rostral ventrolateral medulla (RVLM) (neurons is necessary for a full CVC induced by skin cooling (Ootsuka and McAllen 2005) and by a PGE2-induced fever response (Rathner et al., 2008). We showed that the activity of rRPa neurons accounts for approximately 83% of IL-1β-induced CVC. A minor (~17% of the full CVC response), yet significant component of IL-1β-induced CVC, may thus be attributable to descending excitation of CVC preganglionic neurons from neurons in the RVLM. Activation of the CVC SPmN in the rRPa requires an excitatory input from neurons that likely reside in the POA, but not in the DMH (Rathner et al., 2008, McAllen and McKinley 2018). Our finding that the COX-independent, IL-1β-induced fever involves an increased CVC dependent on the activity of rRPa neurons constitutes a significant advance in our understanding of the mechanisms underlying IL-1-dependent fevers. The source(s) of the excitation driving CVC SPmN during IL-1-dependent fever remains to be investigated.

In contrast to the extensively studied neural circuitry mediating the thermogenic components of COX-2/PGE2-dependent fevers (Morrison and Nakamura 2019), the neural pathways and synaptic integration sites underlying COX-independent fevers are still largely unknown. In the present study, nanoinjections of muscimol in the rRPa prevented or reversed both the BAT thermogenesis and the tachycardia elicited by iv IL-1β. These results indicate that the BAT and cardiac thermogenesis stimulated during the COX-independent fever consequent to iv IL-1β administration, and bacterial, PGE₂-dependent fevers share a dependence on the activation of neurons in the rRPa. These are likely to include the populations of BAT and of cardiac SPmNs residing in rRPa and that project to their respective BAT and cardiac sympathetic preganglionic neuron targets in the spinal cord, which, in turn, excite the sympathetic ganglion cells that innervate BAT and the cardiac sinoatrial node (Morrison and Nakamura 2019). Thus, although the IL-1 receptor is expressed in peripheral sympathetic ganglia (Hart et al., 1993), and IL-1β can excite stellate ganglion cells (Wang et al., 2017), the direct effects of IL-1β in these peripheral ganglia are not sufficient to provoke robust BAT thermogenic, tachycardic or CVC responses in the absence of a supraspinal excitatory drive to sympathetic preganglionic neurons.

Neurons in the DMH provide a direct excitatory input to the rRPa (Kataoka et al., 2014) that is necessary for increasing the BAT metabolism and thermogenesis, as well as the tachycardia induced by injection of PGE2 in the POA or by exposure to a cold ambient temperature (reviewed in (Madden and Morrison 2019, Morrison and Nakamura 2019)). DMH and rRPa neurons are also necessary for COX-independent, stress-induced hyperthermia (Lkhagvasuren et al., 2011). Thus the DMH-RPa pathway appears to be a common output pathway that can be driven by inputs arising from separate brain regions mediating diverse febrile responses, such as cortical regions providing excitatory inputs to the DMH for psychogenic fever (Lkhagvasuren et al., 2011, Kataoka et al., 2020), or preoptic area neurons providing excitatory inputs for COX-2/PGE2-dependent fever (Morrison and Nakamura 2019, da Conceicao et al., 2020, Machado et al., 2020). The brain regions and specific neurons providing the excitatory inputs to the DMH and RPa for IL-1β-induced fever remain to be determined.

Our finding that the thermogenic components of iv IL-1β-evoked fever are prevented by nanoinjection of ionotropic glutamate receptor antagonists in the DMH indicates not only that the activity of thermogenesis-promoting neurons in the DMH is required for iv IL-1β-evoked fever, but also that a glutamatergic excitation of these DMH neurons is critical to maintain their discharge during an iv IL-1β-evoked fever. The sources of the excitatory and inhibitory inputs to thermogenesis-promoting neurons in the DMH mediating IL-1β fever are currently being investigated. Furthermore, investigating the effects of IL-1β in freely moving rats is an important step to fully understand the COX-independent mechanisms triggering inflammatory fever by IL-1β.

5. Conclusion

Our results suggest that IL-1β induces a neuroimmune response mediated by a glutamatergic input to the DMH and a DMH to rRPa sympathoexcitatory pathway for the COX-independent, iv IL-1β-induced BAT thermogenesis and tachycardia, as it is for similar responses during COX-dependent, PGE₂-evoked fever. Differences between the neurocircuitries mediating the BAT thermogenesis during bacterial, COX-dependent fever and during the IL-1β-evoked, COX-independent fever are likely antecedent to the DMH to rRPa axis. Further investigation will be required to determine the sites at which IL-1β acts to trigger COX-independent fevers. A complete picture of the neural pathways that modulate the activity of the different thermoeffectors that control body temperature and energy expenditure in inflammatory fever induced by IL-1β may contribute to our understanding of the mechanisms involved in fevers and inflammatory conditions that are resistant to COX inhibitors, and the identification of novel treatments that avoid the side effects of COX inhibitors.