Hypothalamic TLR2 triggers sickness behavior via a microglia-neuronal axis

Abstract

Various pathophysiologic mechanisms leading to sickness behaviors have been proposed. For example, an inflammatory process in the hypothalamus has been implicated, but the signaling modalities that involve inflammatory mechanisms and neuronal circuit functions are ill-defined. Here, we show that toll-like receptor 2 (TLR2) activation by intracerebroventricular injection of its ligand, Pam3CSK4, triggered hypothalamic inflammation and activation of arcuate nucleus microglia, resulting in altered input organization and increased activity of proopiomelanocortin (POMC) neurons. These animals developed sickness behavior symptoms, including anorexia, hypoactivity, and hyperthermia. Antagonists of nuclear factor kappa B (NF-κB), cyclooxygenase pathway and melanocortin receptors 3/4 reversed the anorexia and body weight loss induced by TLR2 activation. These results unmask an important role of TLR2 in the development of sickness behaviors via stimulation of hypothalamic microglia to promote POMC neuronal activation in association with hypothalamic inflammation.

Figure 1. Centrally activated TLR2 leads to sickness responses.

(A,B) Changes in food intake (A) and body weight (B) were measured in rats for 24 h after icv injection of vehicle (CTL) or Pam3CSK4 (Pam3). The pair-fed group was provided with the average amount of food consumed by Pam3-injected rats for 24 h (n = 6–8 rats/group; *P < 0.05, ***P < 0.0001 by unpaired two-tailed Student’s t-tests). (C,D) Body temperature (C) and locomotor activity (D) were monitored in rats that were treated icv with vehicle or Pam3 using the Vital View system (n = 6–8 rats/group; *P < 0.05, ***P < 0.0001 by unpaired two-tailed Student’s t-tests; NS, not significant). (E,F) Food intake (E) and body weight (F) were determined in TLR2 KO mice for 24 h after icv injection of vehicle or Pam3 (n = 3–4 mice/group, **P < 0.01, ***P < 0.001 by two-way ANOVA; NS, not significant). (G,H) Food intake (G) and body weight (H) were measured in MyD88 KO mice for 24 h after icv injection of vehicle or Pam3 (n = 3–5 mice/group; *P < 0.05, **P < 0.01, ***P < 0.001 by two-way ANOVA; NS, not significant). (A–D) Data from rat models. All data are presented as mean ± s.e.m.

Figure 2. TLR2 signaling-induced morphological change of microglia in the mouse hypothalamic Arc.

Central administration of Pam3CSK4 (Pam3) resulted in microglia-specific induction of TLR2. (A) TLR2 immunosignals (red) were co-localized with signals of Iba-1 (green), a marker for microglia, but not with signals of GFAP (green), a marker for astrocytes, in the arcuate nucleus of the saline-injected mouse brain. (B) Central administration of Pam3 caused exclusive induction of TLR2 in the Iba-1-positive cells. Arrow heads: double-labeled TLR2 and Iba-1-positive cells. (C) Pam3CSK4 (Pam3)-induced activation of microglia and TLR2 expression were absent in the TLR2 KO mice. NC = negative control for TLR2 (without primary antibody). Scale bar = 100 μm (20 μm for higher magnification view in insets).

Figure 3. Microglia activation is responsible for the anorexia and body weight loss induced by TLR2 in the mouse hypothalamic Arc.

(A) Representative images (3–4 mice analyzed) show immunosignals of Iba-1 in the hypothalamic Arc. Scale bar = 100 μm (20 μm for higher magnification view in inset). (B,C) Increased intensity of Iba-1 signals in the Arc (B) of icv Pam3CSK4 (Pam3)-injected mice was significantly attenuated by preadministration of minocycline (Mino), a bacteriostatic antibiotic that effectively inhibits microglia activation. There was no difference in the number of microglia (C) among different treatment groups (CTL, n = 8 section/4 mice; Pam3, n = 6 section/3 mice; Mino + Pam3, n = 8 section/4 mice; ** P < 0.01, ***P < 0.0001 by unpaired two-tailed Student’s t-tests; NS, not significant). (D,E) Pam3-induced anorexia (D) and weight loss (E) were significantly mitigated by administration of Mino for 3 consecutive days prior to icv Pam3 injection. (CTL, n = 5 mice; Pam3, n = 6; Mino, n = 7; Mino + Pam3, n = 7; *P < 0.05, **P < 0.01, ***P < 0.001 by two-way ANOVA; NS, not significant). All data are presented as mean ± s.e.m.

Figure 4. Involvement of NF-κB and COX pathway in the TLR2-associated sickness behaviors.

Mice received an NF-κB inhibitor, Bay 11-7085 (Bay) or rats received a COX inhibitor, indomethacin (Indo), 1 h prior to icv injection of Pam3CSK4 (Pam3), a synthetic TLR2 ligand, and changes in sickness responses were monitored. (A) Hypothalamic content of phosphorylated p65, a subunit of NF-κB, was determined in 8-week-old male rats after icv injection with either vehicle (CTL) or Pam3 (n = 3 rats/group; *P < 0.05 by unpaired two-tailed Student’s t-tests). (B,C) Food intake (B) and body weight (C) were assessed in mice that were treated icv with either vehicle or Bay 1 h prior to Pam3 injection (CTL, n = 4 mice; Pam3, n = 8; Bay, n = 7; Bay + Pam3, n = 4; *P < 0.05, ***P < 0.001 by two-way ANOVA). (D–G) Changes in food intake (D), body weight (E), locomotor activity (F), and body temperature (G) were monitored in rats that were intraperitoneally injected with either vehicle or Indomethacin 1 h prior to administration of Pam3 (n = 4–6 rats/group; **P < 0.01, ***P < 0.001 by two-way ANOVA; NS, not significant). (A,D–G) Data from rat models. (B,C) Mouse models. All data are presented as mean ± s.e.m.

Figure 5. TLR2-induced microglia activation alters synaptic inputs of POMC neurons.

(A) Images of POMC-GFP (green) and Iba1-positive microglia (red) in the arcuate nucleus after saline (CTL) or Pam3CSK4 (Pam3) injection. Scale bar = 100 μm (10 μm in inset). (B) Percent of POMC cells in contact with microglial cells in CTL and Pam3-injected mice (n = 5 section/3 mice/group; *P < 0.05 by unpaired two-tailed Student’s t-tests). (C) Percentage of microglia area of POMC neurons in Pam3- and saline-injected mice (CTL, n = 53 cells/4 mice; Pam3, n = 95 cells/7 mice; ***P < 0.0001 by unpaired two-tailed Student’s t-tests). (D) Triple labeling of POMC neurons (green), microglia (red), and GAD 67 (blue). Scale bar = 10 μm. (E,F) Pam3 decreased percentage of GABAergic synapses on the surface of POMC soma in contact with activated microglia (E), CTL, n = 53 cells/4 mice; Pam3, n = 95 cells/7 mice; ***P < 0.0001 by unpaired two-tailed Student’s t-tests) but not on POMC cells separate from microglia (F), CTL, n = 166 cells/4 mice; Pam3, n = 243 cells/7 mice; NS). (G) Scatter plots show a negative correlation between the GABAergic innervation and the area occupied by microglia onto POMC neurons. (n = 148 cells in contact with microglia/11 mice; r = −0.3269, P < 0.0001 by Pearson correlation coefficient analysis with a two-tailed test). (H) Confocal micrographs of POMC neurons (green), microglia (red), and vGlut2 (blue). Scale bar = 10 μm. (I,J) Glutamatergic synapses on POMC soma in contact with microglia (I), CTL, n = 27 cells/2 mice; Pam3, n = 38 cells/2 mice; ***P < 0.0001 by unpaired two-tailed Student’s t-tests) or separated from microglia (J) CTL, n = 161 cells/2 mice; Pam3, n = 95 cells/2 mice; ***P < 0.0001 by unpaired two-tailed Student’s t-tests). (K) Scatter plots show positive correlation between the glutamatergic innervation and the area occupied by microglia onto POMC neurons. (n = 65 cells in contact with microglia/4 mice; r = 0.6050, P < 0.0001 by Pearson correlation coefficient analysis with a two-tailed test). All data are presented as mean ± s.e.m.

Figure 6. Involvement of melanocortin pathway in the TLR2-induced anorexia and body weight loss.

(A,B) POMC neuronal activity was determined by the change in c-Fos immunoreactivity in the arcuate nucleus of mice that were icv treated with Pam3CSK4 (Pam3) 1 h before sacrifice. Representative photographs (A) and calculated graphs (B) reveal an increase in the number of c-Fos positive POMC cells induced by Pam3 (CTL, n = 4 section/2 mice; Pam3, n = 6 section/3 mice; ***P < 0.0001 by unpaired two-tailed Student’s t-tests). Scale bar = 100 μm. (C–F) Effect of Pam3 on POMC neuronal terminals of the hypothalamic paraventricular nucleus (PVN) was determined by α-melanocyte stimulating hormone (α-MSH) immunoreactivity. Representative images (C) show an increase in α-MSH immunosignals in the PVN after Pam3 injection. Scale bar = 100 μm. Icv administration of Pam3 led an increase in particle number (D), total area (E), and relative density (F) of α-MSH fiber signals (CTL, n = 12 sections/6 mice; Pam3, n = 16 sections/8 mice; *P < 0.05, **P < 0.01 by unpaired two-tailed Student’s t-tests). (G,H) Effect of antagonizing melanocortin 3 and 4 (MC3/4) receptors on Pam3-induced sickness behaviors. Changes in food intake and body weight were monitored in rats after icv treatment with SHU9119, an MC3/4 receptor antagonist. The antagonist completely abolished the inhibitory effect of Pam3 on food intake (G) and body weight (H) (n = 4–5 rats/group; ***P < 0.001 by two-way ANOVA; NS, not significant). (G,H) Data from rat model. All data are presented as mean ± s.e.m.