The glia-derived alarmin IL-33 orchestrates the immune response and promotes recovery following CNS injury

Abstract

Inflammation is a prominent feature of CNS injury that heavily influences neuronal survival, yet the signals that initiate and control it remain poorly understood. Here we identify the nuclear alarmin, interleukin (IL)-33, as an important regulator of the innate immune response after CNS injury. IL-33 is expressed widely throughout the healthy brain and is concentrated in white mater due to predominant expression in post-mitotic oligodendrocytes. IL-33 is released immediately after CNS injury from damaged oligodendrocytes, acting on local astrocytes and microglia to induce chemokines critical for monocyte recruitment. Mice lacking IL-33 have impaired recovery after CNS injury, which is associated with reduced myeloid cell infiltrates and decreased induction of M2 genes at the injury site. These results demonstrate a novel molecular mediator contributing to immune cell recruitment to the injured CNS and may lead to new therapeutic insights in CNS injury and neurodegenerative diseases.

Figure 1.. IL-33 Is Highly Expressed in Myelinated Tissue due to Enrichment in Oligodendrocytes

(A) Mouse peripheral tissues and brain regions were isolated and analyzed for IL-33 transcript (n = 3; one-way ANOVA with Tukey’s multiple comparison test). (B) Representative labeling of OLIG2 and IL-33 in spinal cord gray matter of (top) WT mice (white arrows = OLIG2⁺ IL-33⁺ cells; yellow arrows = OLIG2⁺ IL33⁻ cells) and (bottom) IL-33^−/− mice as a negative control. (C) Representative labeling of CC-1 and IL-33 in WT spinal cord gray matter (white arrows = CC-1⁺ IL-33⁺ cells; yellow arrows = CC-1⁺ IL33⁻ cells; blue arrows = CC-1⁻ IL-33⁺ cells). (D) Quantification of IL-33-, OLIG2-, and CC-1-positive cells in multiple CNS regions (data are presented as mean counts from four to five individual animals). (E) Representative image of OLIG2 and IL-33 labeling in neocortex (white arrows = OLIG2+ IL-33⁺ cells; yellow arrows = OLIG2⁺ IL33⁻ cells; blue arrows = OLIG2⁻ IL-33⁺ cells). (F) Representative labeling of IL-33 and ALDH1L1 in the neocortex (white arrows = ALDH1L1⁺ IL-33⁺ cells; yellow arrows = ALDH1L1⁻ IL33⁺ cells). (G) Quantification of ALDH1L1 and IL-33 labeling in multiple CNS regions (data are presented as mean counts from four to five individual animals). (H) Representative gating strategy for isolating oligodendrocyte lineage cells from whole brain by flow cytometry. Gates for oligodendrocyte antigens were drawn based on fluorescence minus one (FMO) controls, gates for IL-33 were drawn based on IL-33^−/− samples. (I) Histograms showing IL-33 expression in several CNS cell populations isolated from whole brain for flow cytometry. Black lines represent staining of IL-33^−/− brain. (J and K) Summary of MFI of IL-33+ populations (J) and percent of cells expressing IL-33 (K) (n = 3–4; one-way ANOVA with Tukey’s multiple comparison test). Error bars represent mean ± SEM; *p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001.

Figure 2.. IL-33 Is Released after CNS Injury without Induction of New Transcript

(A) Representative image of a spinal cord injury site 1DPI stained for IL-33. IL-33-positive nuclei are present adjacent to the injury (top right) but absent at the injury site (bottom right). Images are representative of four individual animals. (B) CSF was gathered from mice after spinal cord contusion and analyzed for IL-33 by ELISA. IL-33 was detectable only at 1 hr post-injury (n = 3 mice per group). (C) IL-33 ELISA of in vitro contused spinal cords. 4 mm pieces of spinal cord were contused in individual wells of a 96-well plate and incubated for 10 min, 1 hr, or 30 hr before supernatants were collected for ELISA (n = 5 individual animals; one-way ANOVA with Tukey’s multiple comparison test). (D) Spinal cord segments release more IL-33 with contusion than without (n = 5 individual animals; paired two-tailed t test). (E) In vitro contused spinal cords were harvested and analyzed for IL-33 transcript, showing no induction of IL-33 mRNA (n = 5 mice per group; one-way ANOVA with Tukey’s multiple comparisons test). Error bars represent mean ± SEM; *p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001.

Figure 3.. IL-33 Is Beneficial after CNS Injury and IL-33^−/− Have Decreased M2 Macrophage Activity at the Injury Site

(A) Representative images of Fluorogold-labeled RGCs with and without optic nerve crush in WT and IL-33^−/−. (B) Quantification of RGC survival in WT and IL-33^−/− animals (n = 9–11; two-tailed Student’s t test). (C) Representative images of Brn3a-labeled RGCs with and without glutamate treatment. (D) Quantification of RGC survival after glutamate toxicity (n = 5; two-tailed Student’s t test). (E) Recovery after spinal cord contusion (90 KDyn) as measured by the Basso Mouse Scale (BMS) (Basso et al., 2006) (n = 8 mice per group, representative of 3 experiments with 70–90 KDyn severity; repeated-measures two-way ANOVA with Sidak’s multiple comparisons test). Motor coordination was assessed with the (F) Rotarod and (G) Pole tests. (F) After two training trials, WT and IL-33^−/− mice were placed on an accelerating rod and the latency to fall was recorded (n = 7–10; two-tailed Student’s t test). (G) After one training trial, mice were placed at the top of a narrow pole and latency to descend the pole was recorded (n = 9; two-tailed Student’s t test). (H) Representative injury sites of WT and IL-33^−/− animals 14 days after spinal cord contusion labeled with GFAP and IBA1. (I) Volume of lesion sites at 23DPI as measured by GFAP delineation. (J) 1 cm of spinal cord tissue centered on the 1DPI lesion site was analyzed for transcript levels of M2 genes in WT and IL-33^−/− animals (n = 3, representative of two experiments; multiple two-tailed Student’s t tests). (K) BMDMs were treated with IL-4 (10 ng/ml), anti-ST2 antibodies (10 ng/ml) and dissected 2.5 mm segments of mouse spinal cord for 24 hr in vitro before lysates were analyzed for Arginase 1 transcript (n = 3; (main graph) one-way ANOVA with Tukey’s multiple comparisons test; (inset) two-tailed Student’s t test). (L) BMDMs were treated with IL-4 (10 ng/ml), anti-ST2 antibodies (10 ng/ml), and optic nerves from WT or IL-33^−/− animals (1 optic nerve/well of a 24-well plate, cut into 2 pieces) for 24 hr and analyzed for Arg1 transcript (n = 3, representative of 2 experiments; (main graph) one-way ANOVA with Tukey’s multiple comparisons test; (inset) two-tailed Student’s t test). (M and N) WT or ST2^−/− BMDMs were treated with IL-4 (10 ng/ml), IFNγ (100 ng/ml), or optic nerves (as in B), for 24 hr and analyzed for Arg1 or Nos2 transcript (n = 3, representative of two experiments; two-way ANOVA with Sidak’s multiple comparisons test). Error bars represent mean ± SEM; *p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001.

Figure 4.. IL-33^−/− Mice Have Impaired Myeloid Cell Recruitment and Chemokine Induction after CNS Injury

(A) Example of gating for peripheral myeloid cells at the injury site. Cells were gated as follows: Hoescht⁺, CD45⁺, Singlets, live cells, CD11b⁺, CD45^hi. (B–E) Flow cytometry analysis of the injury site 1 and 7 days after spinal cord injury. (B) Representative plots of WT and IL-33^−/− sites at 1DPI. (C) Mean cell numbers at the injury site of total peripheral myeloid cells (CD11b⁺/CD45^hi) and subsets of monocytes (Ly6C^hi/Ly6G^lo), granulocytes (Ly6C^int/Ly6G⁺), and macrophages (Ly6C^lo/Ly6G^lo) (n = 3 mice per group, representative of two experiments; two-way ANOVA with Sidak’s multiple comparison’s test). (D) Representative flow plots and (E) summarized results comparing WT and IL-33^−/− injury sites at 7DPI. (n = 3 mice per group; two-way ANOVA with Sidak’s multiple comparisons test). (F) Representative images of Arg1+ cells at the lesion site 1DPI. (G) 1DPI WT and IL-33^−/− injury sites were analyzed by immunofluorescence for Arg1⁺ cell number. Spinal cords were sliced longitudinally and collected at equivalent intervals in all mice, and at least seven equivalent slices were analyzed per mouse. (n = 3; two-tailed Student’s t test). (H) Luminex chemokine analysis of mixed glial cultures after treatment with IL-33 (100 ng/ml) for 24 hr. Data are shown as fold induction relative to untreated cultures (n = 4; multiple two-tailed Student’s t test). (I) Chemokine transcript levels from the injury site at 1DPI in WT versus IL-33^−/− mice (n = 3 mice per group, representative of two experiments; multiple two-tailed Student’s t test). (J) qRT-PCR analysis of chemokine induction from sorted CD11b⁺ and CD11b⁻ mixed glial culture populations treated with IL-33 (100 ng/ml) overnight. Data are shown as fold induction of (Ji) Ccl2 and (Jii) Cxcl10 relative to untreated (n = 3; one-way ANOVA with Tukey’s multiple comparisons test). (K) qRT-PCR analysis of ST2 transcript levels from CD11b⁺ and CD11b⁻ sorted from the injury site (n = 3; two-way ANOVA with Sidak’s multiple comparisons test). (L) CCL2 treatment (10 ng/mouse) recovery deficits in IL-33^−/− animals (90 KDyn) (n = 5–6; repeated-measures two-way ANOVA). Error bars represent mean ± SEM; *p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001).