Obesity-associated microglial inflammatory activation paradoxically improves glucose tolerance

Abstract

Hypothalamic gliosis associated with high-fat diet (HFD) feeding increases susceptibility to hyperphagia and weight gain. However, the body-weight-independent contribution of microglia to glucose regulation has not been determined. Here, we show that reducing microglial nuclear factor κB (NF-κB) signaling via cell-specific IKKβ deletion exacerbates HFD-induced glucose intolerance despite reducing body weight and adiposity. Conversely, two genetic approaches to increase microglial pro-inflammatory signaling (deletion of an NF-κB pathway inhibitor and chemogenetic activation through a modified Gq-coupled muscarinic receptor) improved glucose tolerance independently of diet in both lean and obese rodents. Microglial regulation of glucose homeostasis involves a tumor necrosis factor alpha (TNF-α)-dependent mechanism that increases activation of pro-opiomelanocortin (POMC) and other hypothalamic glucose-sensing neurons, ultimately leading to a marked amplification of first-phase insulin secretion via a parasympathetic pathway. Overall, these data indicate that microglia regulate glucose homeostasis in a body-weight-independent manner, an unexpected mechanism that limits the deterioration of glucose tolerance associated with obesity.

Keywords: POMC; TNF; chemogenetic; glucose sensing; glucose tolerance; hypothalamus; insulin; microglia; obesity; parasympathetic.

Figure 1.. Loss of microglial IKKβ dampens inflammatory signaling and reduces HFD-induced weight gain but does not improve glucose tolerance.

(A) Inflammatory cytokine expression in primary microglial cultures from IKKβ-MGKO and control (Ctl) mice treated 6-h with 100 ng/ml lipopolysaccharide (LPS). N=3-4 wells/genotype/treatment. (B) Body weight and (C) fat mass after 10 weeks of ad libitum CD and HFD feeding. N=10-12/group. (D) GTT and (E) area under the curve (AUC) at 10 weeks of CD or HFD. N=10-12/group. (F-G) AUC from E versus 10 week weight gain in IKKβ-MGKO (red) and Ctl (gray) mice. All values are mean ± SEM. (A-C) two-way ANOVA with Bonferroni post-hoc test. (D-E) Student’s t-test. (F-G) Linear regression. * p < 0.05, *** p < 0.001, **** p <0.0001.

Figure 2.. Microglial IKKβ deletion exacerbates HFD-induced glucose intolerance and insulin resistance in weight-matched mice.

(A) Body weight of ad libitum HFD-fed IKKβ-MGKO (IKKβ-MGKO AL) and HFD pair-fed control mice (Ctl PF). N=7/group. (B) Gonadal fat pad weights after 14 weeks of HFD. (C) GTT and (D) AUC after 7 weeks of HFD. (E) Glucose-stimulated insulin secretion (2 g/kg glucose; gavage). (F) Insulin (brown) immunostaining after 10 week of HFD. Scale bar = 500 μm. (G) Quantification of F. N=4/group. (H) 16-h fasted blood glucose. (I) Liver mRNA expression (relative to Ctl PF group) of glucoregulatory enzymes in overnight fasted mice after 14 weeks of HFD. N=8-10/group. (J) ITT and (K) Area-over-curve (AOC) at 10 weeks HFD. (L) pAkt (Ser473) and total Akt in tissue homogenates from liver, skeletal muscle, and epididymal white adipose tissue (eWAT) from 16-h fasted mice given saline (−) or insulin (+) (5 U/kg, i.p., 15 minutes) after 14 weeks on HFD. (M-O) Densitometric quantification of L. N=4-6/group. All values are mean ± SEM. (A-B, J, M-O) Two-way ANOVA with Bonferroni post-hoc test. (C-E, G-I, K) Student’s t-test. * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001.

Figure 3.. Hypothalamic glucoresponsiveness is impaired by loss of microglial IKKβ in obese mice.

(A) Hypothalamic mRNA levels (relative to Ctl PF group) of glucose sensing genes after 16 weeks of HFD. N=7/group. (B) Hypothalamic c-Fos (red) and DAPI (blue) immunostaining 1 hour after glucose administration (2 g/kg, i.p.) in 10 week HFD-fed mice. Arcuate nucleus, ARC; ventromedial hypothalamus, VMH; dorsomedial hypothalamus, DMH; lateral hypothalamic area, LHA. Scale bar = 100 μm. (C) Quantification of B. N=3-4/group. All values are mean ± SEM. (A) Student’s t-test. (C) Two-way ANOVA with Bonferroni post-hoc test. * p < 0.05, ** p < 0.01.

Figure 4.. Derepression of microglial IKKβ/NF-κB signaling through A20 (Tnfaip3) deletion improves glucose tolerance.

(A) Schematic diagram: Mice received lethal irradiation with head shielding to protect microglia and preserve blood-brain barrier integrity. Subsequently, adoptive transfer of Tnfaip3^fl/fl bone marrow yielded Cre⁻ Tnfaip3^fl/fl controls (A20-BMT Control) and Cx3Crl^CreER::Tnfaip3^fl/fl (A20-BMT-MGKO) chimeric mice with Cre⁻ peripheral myeloid cells and Cre⁺ microglia. Timeline: After 6 weeks recovery, mice received 6 weeks CD or HFD followed by tamoxifen (TMX) to induce microglial A20 deletion. Studies B-H were performed 1 week after TMX. (B) Hypothalamic TNF (green) and Iba1 (red) immunostaining from CD-fed BMT mice. Scale bars = 100 μm. (C) Quantification of B. N=7/group, 6 sections/mouse. (D) Body weight of CD-fed mice. N=6/group. (E) GTT and (F) AUC in CD-fed mice. (G) GTT and (H) AUC in HFD-fed mice. All values are mean ± SEM. (C-D, F, H) Student’s t-test. (E, G) Two-way ANOVA. ** p < 0.01, *** p < 0.001, **** p < 0.0001.

Figure 5.. Acute pharmacogenetic activation of microglia improves glucose homeostasis in CD and HFD-fed mice.

(A-M) hM3D mice and littermate controls were fed ad libitum CD. (A) GTT and (B) AUC with systemic CNO (1 mg/kg, i.p.). N=11-14/group. (C) GTT and (D) AUC with high-dose systemic CNO (5 mg/kg, i.p.), N=10-11/group. (E-I) Mice were maintained on CNO drinking water (25 μg/mL) for 14 days then returned to normal water. N=9/group. (E) Body weights during treatment. Arrows indicate dates of GTTs in F and H. (F) GTT and (G) AUC following 2 days of CNO water. (H) GTT and (I) AUC after 7 days of normal water. (J) GTT and (K) AUC, and (L) ITT and (M) area over the curve (AOC) after central CNO treatment (1μg, i. c.v.), N=8/group. (N-R) hM3D mice and littermate controls were fed ad libitum HFD. (N) Body weights over 10 weeks of HFD. (O) GTT and (P) AUC after 4 weeks HFD with central CNO (1 μg, i.c.v.), N=8-10/group. (Q) GTT and (R) AUC after 8 weeks HFD with systemic CNO (5 mg/kg, i.p.), N=11-14/group. All values are mean ± SEM. (A, C, F, H, J, L, O, Q) Two-way ANOVA. (B, D, G, I, K, M, P, R) Student’s t-test. * p < 0.05, ** p < 0.01, *** p < 0.001.

Figure 6.. Microglial regulation of glucose homeostasis involves central TNF and melanocortin signaling.

(A-B) GTT of mice administered systemic CNO (1 mg/kg, i.p.) and (A) vehicle (IgG-Fc) or (B) TNF neutralizing antibody etanercept (TNFnab) (2 μg, i.c.v.), N=5-7/group. (C) AUC of A and B. (D) GTT and (E) AUC in wild-type mice administered TNFα (2 pmol, i.c.v.) or vehicle, N=8/group. (F-I) CD-fed control and hM3D mice injected with CNO (5 mg/kg, i.p.) followed at 1 hour intervals by TNFnab (2 μg, i.c.v.) or saline vehicle, glucose (2 g/kg, i.p.), and perfusion for IHC analysis. (F) Hypothalamic immunostaining of c-Fos (yellow) and DAPI (blue). Arcuate nucleus, ARC; ventromedial hypothalamus, VMH; dorsomedial hypothalamus, DMH; lateral hypothalamic area, LHA. Scale bars = 100 μm. (G) Quantification of F. N=5-8/group, 1-6 sections/mouse. (H) Hypothalamic immunostaining of c-Fos (red) and β-endorphin (green; representing POMC neurons) in the ARC. White arrowheads indicate double-positive cells. Scale bars = 50 μm. (I) Quantification of c-Fos⁺ cells as a percent of β-endorphin cells from H. N=5-8/group, 4 sections/mouse. (J) GTT and (K) AUC in hM3D and control mice administered intranasal HS014 (50 μg/naris) or saline and low-dose CNO (0.5 mg/kg, i.p.), N=6-7/group. All values are mean ± SEM. (C, K, I) Two-way ANOVA with Šidák post-hoc test. (E) Student’s t-test. (G) Mixed model with Tukey post-hoc test. * p < 0.05, ** p < 0.01.

Figure 7.. Chemogenetic microglial activation increases first-phase insulin secretion via a parasympathetically-mediated process.

(A-E) Frequently sampled intravenous glucose tolerance testing (fsIVGTTs) was performed on 12-week HFD-fed control and hM3D mice treated with systemic CNO (5 mg/kg, i.p.). (A) fsIVGTT glucose and (B) insulin with AUC insets. (C-E) Minimal model parameters calculated from fsIVGTT glucose and insulin data, N=3-5/group. Acute insulin response to glucose, AIRg. (F-K) hM3D mice and littermate controls were fed ad libitum chow. (F) GTT and (G) AUC after atropine (3 mg/kg, i.p.) or saline and systemic CNO (0.5 mg/kg, i.p.), N=7-8/group. (H-I) Effect of atropine (3 mg/kg, i.p.) or saline and systemic CNO (0.5 mg/kg, i.p.) on (H) glucose and (I) insulin levels at 0 and 8 minutes post-glucose injection (2 g/kg, i.p.), N=9-11/group. (J) Hindbrain immunostaining of choline acetyltransferase (ChAT; green) and c-Fos (red) after systemic CNO (5 mg/kg, i.p.). Area postrema, AP; nucleus tractus solitarius, NTS; dorsal motor nucleus of the vagus, DMV. Scale bars = 100 μm. (K) Quantification of J, N=5/group, 2-4 sections/mouse. All values are mean ± SEM. (A, B, K) Student’s t-test. (C-E) Welch’s one-sided t-test. (G, I) Two-way/three-way ANOVA with Sidak post-hoc test. * p < 0.05, ** p < 0.01, *** p < 0.001.