Inflammatory Resolution Triggers a Prolonged Phase of Immune Suppression through COX-1/mPGES-1-Derived Prostaglandin E2

Abstract

Acute inflammation is characterized by granulocyte infiltration followed by efferocytosing mononuclear phagocytes, which pave the way for inflammatory resolution. Until now, it was believed that resolution then leads back to homeostasis, the physiological state tissues experience before inflammation occurred. However, we discovered that resolution triggered a prolonged phase of immune suppression mediated by prostanoids. Specifically, once inflammation was switched off, natural killer cells, secreting interferon γ (IFNγ), infiltrated the post-inflamed site. IFNγ upregulated microsomal prostaglandin E synthase-1 (mPGES-1) alongside cyclo-oxygenase (COX-1) within macrophage populations, resulting in sustained prostaglandin (PG)E₂ biosynthesis. Whereas PGE₂ suppressed local innate immunity to bacterial infection, it also inhibited lymphocyte function and generated myeloid-derived suppressor cells, the net effect of which was impaired uptake/presentation of exogenous antigens. Therefore, we have defined a sequence of post-resolution events that dampens the propensity to develop autoimmune responses to endogenous antigens at the cost of local tissue infection.

Keywords: adaptive homeostasis; autoimmunity; eicosanoid; immune scarring; macrophage.

Graphical abstract

Figure 1

IFNγ-Induced IP-10/CXCL10 Triggers Post-resolution Monocyte Infiltration Wild-type mice had 0.1 mg zymosan injected into their peritoneal cavity with the cellular infiltrate analyzed by polychromatic flow cytometry starting from when inflammation typically resolves in this model. (A–D and G) Data show the accumulation of (A) NK cells followed by cell-free inflammatory exudates levels of (B) IFNγ, (C) MIG, (D) IP-10, and (G) MCP-1. (E and F) The key post-resolution cell types expressing IFN (E), and the intracellular staining for IFNγ in these cells at days 9/14 after zymosan injection (F). (H and I) Panels in (H) show the numbers of monocytes in MIIG mice (animals whose macrophages are insensitive to IFNγ) at day 14 after zymosan, whereas (I) confirms that IP-10 only is responsible for the infiltration of post-resolution monocytes in this model. ^∗p ≤ 0.05; ^∗∗p ≤ 0.01. Data are expressed as mean ± SEM; n = 5 mice/group.

Figure 2

Lipidomic Profiling of Inflamed and Post-resolved Tissues (A) Peritoneal cell-free inflammatory exudates from mice that received 0.1 mg zymosan were analyzed by LC-MS/MS at indicated time points. (B) Total peritoneal cells were subjected to western blotting to determine the temporal expression of the prostanoid-generating enzyme cascade. (C) The profile of mononuclear phagocytes in the naive and post-inflamed cavity. (D) The relative proportions of tissue-resident macrophages, which were labeled positively with PkH-PCL^red when injected into the naive peritoneum versus infiltrating monocyte-derived macrophages, which are cell tracker negative. (E) FACS was used to separate tissue-resident macrophages (TR-Mϕ^naive), tissue-resident macrophages that experience inflammation (TR-Mϕ^inflam.), infiltrated Ly6c^hi/lo monocytes, and infiltrating monocyte-derived macrophages (mo-Mϕ) to determine cell expression of COXs and their downstream synthase. Data are expressed as mean ± SEM; n = 5 mice/group.

Figure 3

Post-resolution EP Receptor Expression (A) Total peritoneal cells were subjected to western blotting to determine the temporal expression of prostanoid receptors. (B and C) In addition to (B) monocyte/macrophage populations, the post-resolution infiltration of (C) CD4⁺, CD8⁺, and CD19⁺ lymphocytes were FACS sorted to determine EP expression levels on individual post-resolution myeloid and lymphoid populations. Data are presented as mean ± SEM; n = 6 mice per group.

Figure 4

IFNγ Triggers Post-resolution mPGES-1 Expression (A and B) The effects of IFNγ as well as factors it triggers, including MIG and IP-10 (used at levels found at day 9 in the peritoneum), on macrophage expression of COX and its downstream synthase isoform expression (A), and (B) shows quantification for mPGES-1. (C and D) Zymosan was injected into macrophages insensitive to IFNγ (MIIG) mice, from which post-resolution macrophage populations were FACS sorted followed by qPCR to determine (C) mPGES-1 and (D) mPGES-2 expression levels. (E–H) T cells were isolated from the peritoneum at day 14 post-zymosan and incubated ex vivo with PGE₂ (E), while the effects of an EP4 agonist on cell-free exudate levels of IFNγ (F), MIG (G), and IP-10 (H) was determined at day 21. ^∗p ≤ 0.05; ^∗∗p ≤ 0.01. Data are expressed as mean ± SEM; n = 5 mice/group.

Figure 5

Post-resolution Tissues Are in a State of PGE₂-Mediated Innate Immune Suppression Live bacteria (St. pneumonia) were injected into either naive mice or mice bearing a 0.1-mg-zymosan-induced peritonitis at day 21. Separate groups of zymosan-injected mice were dosed from day 6 post-zymosan injection with either MF-498 (EP4 antagonist) or naproxen for two weeks. (A) How mice over time became progressively sick following bacteria. This was assessed using a “murine sickness score,” which was developed in association with the UCL animal welfare group and veterinary surgeon; see Experimental Procedures. (B) The number of surviving bacteria in the blood of these animals. ^∗p ≤ 0.05; ^∗∗p ≤ 0.01. Data are expressed as mean ± SEM; n = 6 mice/group.

Figure 6

Post-resolution PGE₂ Inhibits Adaptive Immunity (A and B) CD4⁺ (A) and CD19⁺ (B) cells were FACS sorted from the post-resolving cavity (day 14) and incubated with increasing concentrations of PGE₂ equivalent to that found in the peritoneum at the same time (see Figure 2A), with/without EP receptor antagonists. (C and D) The (C) impact of dosing mice with an EP4 receptor antagonist (MF-498 from day 6 until day 21) on T cell numbers in situ as well as their (D) phenotype as determined by intracellular IFNγ. (E and F) Inhibiting PGE₂ synthesis (E) reduces numbers of myeloid-derived suppressor cells (sub-panel C) while increasing numbers of (F) dendritic cells in the peritoneum at day 14. (G) Taking this further, methylated BSA (mBSA) was injected into the peritoneum or naive mice as well as those bearing a 0.1 mg zymosan at day 14; controls for this experiment were wild-type mice sensitized with complete Freund’s adjuvant containing mBSA with recall assays carried out on T/B cells. ^∗p ≤ 0.05; ^∗∗p ≤ 0.01; ^∗∗∗p ≤ 0.001. Data are expressed as mean ± SEM; n = 6 mice/group.

Figure 7

Autoantibodies Generated by Inflammation Driven by 10 mg Zymosan Are Inhibited by PGE₂ (A) Wild-type mice were injected with 10 mg zymosan and subjected to polychromatic flow cytometry at the indicated time points for the determination of cells of the monocyte/macrophage lineage. (B) Following the injection of cell-tracker dyes (PKH red) into the naive peritoneum, we were able to discern tissue-resident from infiltrating monocyte-derived macrophages (mo-Mϕ). (C and D) Levels of the (C) classic monocyte chemoattractant MCP-1 were determined in the peritoneal fluid, whereas at day 14 post-zymosan, (D) tissue-resident from infiltrating monocyte-derived macrophages were FACS sorted for the determination of their phenotype by PCR. (E–G) At this time point, profiles of (E) apoptotic lymphocytes and granulocytes are shown alongside (F) serum levels of antibodies to dsDNA occurring in the absence of (G) peritoneal PGE₂, effects that were reversed when (H) mice receiving 10 mg zymosan were dosed from days 6 to 21 with a stable PGE₂ analog. Data are expressed as mean ± SEM; n = 6 mice/group.

Figure 7

Autoantibodies Generated by Inflammation Driven by 10 mg Zymosan Are Inhibited by PGE₂ (A) Wild-type mice were injected with 10 mg zymosan and subjected to polychromatic flow cytometry at the indicated time points for the determination of cells of the monocyte/macrophage lineage. (B) Following the injection of cell-tracker dyes (PKH red) into the naive peritoneum, we were able to discern tissue-resident from infiltrating monocyte-derived macrophages (mo-Mϕ). (C and D) Levels of the (C) classic monocyte chemoattractant MCP-1 were determined in the peritoneal fluid, whereas at day 14 post-zymosan, (D) tissue-resident from infiltrating monocyte-derived macrophages were FACS sorted for the determination of their phenotype by PCR. (E–G) At this time point, profiles of (E) apoptotic lymphocytes and granulocytes are shown alongside (F) serum levels of antibodies to dsDNA occurring in the absence of (G) peritoneal PGE₂, effects that were reversed when (H) mice receiving 10 mg zymosan were dosed from days 6 to 21 with a stable PGE₂ analog. Data are expressed as mean ± SEM; n = 6 mice/group.