MAFB prevents excess inflammation after ischemic stroke by accelerating clearance of damage signals through MSR1

Abstract

Damage-associated molecular patterns (DAMPs) trigger sterile inflammation after tissue injury, but the mechanisms underlying the resolution of inflammation remain unclear. In this study, we demonstrate that common DAMPs, such as high-mobility-group box 1 (HMGB1), peroxiredoxins (PRXs), and S100A8 and S100A9, were internalized through the class A scavenger receptors MSR1 and MARCO in vitro. In ischemic murine brain, DAMP internalization was largely mediated by MSR1. An elevation of MSR1 levels in infiltrating myeloid cells observed 3 d after experimental stroke was dependent on the transcription factor Mafb. Combined deficiency for Msr1 and Marco, or for Mafb alone, in infiltrating myeloid cells caused impaired clearance of DAMPs, more severe inflammation, and exacerbated neuronal injury in a murine model of ischemic stroke. The retinoic acid receptor (RAR) agonist Am80 increased the expression of Mafb, thereby enhancing MSR1 expression. Am80 exhibited therapeutic efficacy when administered, even at 24 h after the onset of experimental stroke. Our findings uncover cellular mechanisms contributing to DAMP clearance in resolution of the sterile inflammation triggered by tissue injury.

Main

Ischemic stroke, a common cause of severe disability and death worldwide, is typically associated with acute sterile inflammation1,2. Very early thrombolysis and mechanical thrombectomy have been shown to improve outcome after ischemic stroke; nevertheless, additional therapies are urgently needed, as most patients with ischemic stroke receive insufficient benefit from existing therapies owing to their limited clinical indications3. Recently, post-ischemic inflammation, which lasts about a week after stroke onset, has been considered as a potential target in widening the therapeutic timeframe1,3,4. On the other hand, extension of endogenous mechanisms for resolution of the inflammation associated with infection or atherosclerosis has been proposed to reduce tissue injury5,6. As the resolution of inflammation is often associated with tissue repair and regeneration7,8,9, understanding the molecular mechanisms underlying the resolution of cerebral post-ischemic inflammation could aid in establishing novel therapies for ischemic stroke.

The extracellular release of DAMPs from dead cells is an important trigger of sterile inflammation after tissue injury10,11,12,13. HMGB1, the S100A8 and S100A9 (S100A8/A9) proteins, and the PRX family of proteins are well-characterized DAMPs that have been shown to be involved in central nervous system (CNS) injury. HMGB1 is an early DAMP released within a few hours of stroke onset14,15 that disrupts the blood–brain barrier16,17. The S100A8/A9 proteins are produced by infiltrating macrophages in the ischemic brain18 and play important roles in promoting inflammation19,20. The PRX family of proteins is also known to be involved in inflammatory response to various tissue injuries21,22,23,24,25,26,27. Although increased expression of PRXs within brain cells exerts neuroprotection during ischemic stress28, PRXs released extracellularly from dead cells activate Toll-like receptor 2 (TLR2) and TLR4 in immune cells23,24,25,26,29 and trigger the production of inflammatory cytokines, including IL-1β and IL-23, which induce T cell–mediated delayed inflammation after ischemic stroke25,30,31.

Acute inflammatory responses triggered by DAMPs exaggerate tissue injury and eventually lead to further release of DAMPs by promoting cell death10. We therefore hypothesized that clearance of DAMPs from injured tissue is required for resolution of inflammation. In this study, we identified Msr1, Marco, and Mafb as genes with key roles in the internalization of DAMPs by mononuclear phagocytes. We propose a new strategy to improve ischemic brain injury by enhancing DAMP clearance through MAFB-mediated induction of MSR1.

Results

DAMPs are efficiently removed by infiltrating mononuclear phagocytes

To clarify the fate of DAMPs after inflammatory responses, we incubated cells isolated from day 3 post-ischemic mouse brain, which was generated by transient middle cerebral artery occlusion (MCAO) with DyLight-fluorescence-labeled recombinant PRXs, HMGB1, and S100A8/A9 proteins (F-PRXs, F-HMGB1, and F-S100A8/A9, respectively; Supplementary Fig. 1a). Microscopic analysis revealed that F-PRXs, F-HMGB1, and F-S100A8/A9 were internalized by F4/80+ myeloid cells collected from ischemic brain but did not colocalize with F4/80 on the cell membrane (Fig. 1a). F-PRXs, F-HMGB1, and F-S100A8/A9 were diffusely localized in the cytoplasm after 1 h of incubation and formed vesicular structures within 6 h in almost all F4/80+ myeloid cells (Fig. 1b). The vesicle-like structures were co-stained with a lysosome marker, LysoTracker (Fig. 1c), suggesting that PRXs, HMGB1, and S100A8/A9 are internalized by F4/80+ myeloid cells for lysosomal degradation.

Figure 1: PRXs, HMGB1, and S100A8/A9 are internalized by infiltrating mononuclear phagocytes.
figure1

(a) Microscopic images of the internalization (after 1 h of exposure) of F-PRXs (a mixture of PRX1, PRX2, PRX5, and PRX6), F-HMGB1, and F-S100A8/A9 (a mixture of S100A8 and S100A9) by F4/80+ myeloid cells collected from day 3 post-ischemic brain. Scale bars, 10 μm. (b,c) Microscopic images of F-PRX, F-HMGB1, and F-S100A8/A9 proteins internalized by cells collected from day 3 post-ischemic brain (F4/80+ cells are shown; scale bars, 10 μm). (b) Internalized F-DAMPs are diffusely located in the cytoplasm (after 1 h of incubation) and then localize to vesicle-like structures (after 6 h of incubation). Arrowheads indicate vesicles containing internalized F-PRXs, F-HMGB1, or F-S100A8/A9. (c) Vesicles were co-stained with LysoTracker Red after 6 h of incubation. (d) The internalization (after 1 h of exposure) of F-GST, F-PRXs, F-HMGB1, and F-S100A8/A9 by cells collected from day 3 post-ischemic brain was analyzed by FACS (n = 6 mice per group). CD45hiCD11bhi, CD45intCD11bint, and CD45 populations were individually separated by FACS, as shown to the left. (e) Gr-1hi and Gr-1lo cells were individually separated from the CD45hiCD11bhi population by FACS and examined for the internalization of F-PRXs. In d,e, the fluorescence intensity of the cells is plotted in arbitrary units. **P < 0.01, ***P < 0.001, one-way ANOVA with Dunnett's correction. Data are shown as mean ± s.e.m. Data points (circles) overlaid upon the group means in panel (d) represent individual values from animal.

We next sought to identify the cell types responsible for DAMP internalization after transient MCAO. Flow cytometric analysis revealed that the CD45hiCD11bhi population efficiently internalized F-PRXs, F-HMGB1, and F-S100A8/A9 but not a control protein (fluorescence-conjugated glutathione S-transferase, F-GST) (Fig. 1d). In contrast, fluorescence-conjugated DAMPs were not efficiently internalized by the CD45intCD11bint population, which includes microglia32,33,34, or the CD45 population, which includes non-hematopoietic cells (Fig. 1d). After MCAO, the CD45hiCD11bhi population includes Gr-1hi neutrophils and F4/80+Gr-1lo mononuclear phagocytes, which comprise infiltrated or CNS-associated macrophages, monocyte-derived cells, and activated microglia12,13,32. We found that Gr-1lo mononuclear phagocytes internalized F-PRXs efficiently, whereas Gr-1hi cells did not (Fig. 1e). The majority of the F4/80+Gr-1lo mononuclear phagocytes in the CD45hiCD11bhi population were derived from transplanted bone marrow (BM) cells in chimeric mice head-shielded during irradiation (Supplementary Fig. 1b), indicating that PRXs, HMGB1, and S100A8/A9 were selectively internalized by infiltrating mononuclear phagocytes.

Msr1 and Mafb are critical for the clearance of DAMPs in vitro

To identify the molecular mechanisms involved in the selective internalization of DAMPs, we started with RAW264.7 cells, a macrophage-like cell line that efficiently internalizes F-PRXs (Supplementary Fig. 2a), and established mutant RAW264.7 clones that were deficient in the internalization of F-PRXs. Parental RAW264.7 cells were treated with ENU (N-ethyl-N-nitrosourea), which randomly introduces DNA mutations35 (Supplementary Fig. 2a,b). Cells that failed to internalize F-PRXs were repeatedly sorted by FACS, and, through subsequent limiting dilution, we established three mutant cell lines (Mut 1–3) that lacked the ability to internalize F-PRXs (Fig. 2a and Supplementary Fig. 2b–d).

Figure 2: Identification of Msr1 and Mafb as essential genes for the internalization of DAMPs.
figure2

(a) F-PRX internalization by RAW264.7 cells (RAW) and ENU-induced mutant RAW264.7 lines (Mut 1–3) co-stained with F4/80. Scale bars, 10 μm. (b) Heat map of microarray analysis results showing transmembrane receptor and transcription regulator genes whose expression was deficient in mutant RAW264.7 clones in comparison to parental RAW264.7 cells. The gene expression measurements, as determined by microarray signal after global normalization, are shown in the heat map. (c,d) Internalization of F-PRXs (c) and F-HMGB1 or F-S100A8/A9 (d) by mutant RAW264.7 Mut 2 cells expressing the indicated transmembrane receptors or transcription regulators via lentiviral transduction (n = 3 independent cell preparations from 4 independent experiments). MFI, mean fluorescence intensity. (e) Histogram of MSR1 and MARCO expression levels (arbitrary units) in mock-transduced and Mafb-overexpressing mutant RAW264.7 clone (Mut 2) and parental RAW cells. An IgG isotype control was used to measure background fluorescence as a negative control (n = 5 independent cell preparations from 3 independent experiments). (f) Luciferase reporter assays of the Msr1 gene promoter region using the indicated deletion and mutant constructs. The conserved DNA sequences of two putative MAREs are shown above (black). Luciferase activities were compared between mock- and Mafb-transfected HEK293 cells (n = 3 independent cell preparations from 3 independent experiments). (g) ChIP assays of the two putative MAREs in the Msr1 gene promoter region using mock-transduced and Mafb-overexpressing mutant RAW264.7 (Mut 2) lines (n = 3 independent cell preparations from 3 independent experiments). *P < 0.05, ***P < 0.001 versus mock-treated cells (ce,g) or activity for Mafb-transfected −536/Luc cells (f) (c,d,f, one-way ANOVA with Dunnett's correction; e,g, two-sided Student's t-test). Data are shown as mean ± s.e.m.

We next compared the gene expression profiles of parental RAW264.7 cells and Mut 1–3 cells. To analyze deficient genes encoding receptors and transcription regulators, we selected genes whose expression in Mut 1–3 cells was less than 25% of that in parental RAW264.7 cells (Fig. 2b). The cDNAs of these genes were cloned into a lentiviral vector and reintroduced into the mutant lines, and the internalization of F-PRXs was then examined. Among these genes, overexpression of Msr1 and Mafb was found to be sufficient for efficient internalization of F-PRXs (Fig. 2c). As MSR1 belongs to the class A scavenger receptor family36, we examined other scavenger receptors in this family, including Fcrls (Msr2), Marco, and Scara3, whose encoded proteins are structurally related to MSR1. Although Marco expression was absent in parental RAW264.7 cells, forced expression of Marco, but not forced expression of Msr2 or Scara3, also restored the internalization of F-PRXs (Fig. 2c). We confirmed an essential role of Msr1 in F-PRX internalization by disrupting the gene using the CRISPR–Cas9 system (Supplementary Fig. 2e). Furthermore, Msr1 and Marco were cloned from the lentiviral cDNA expression library constructed from the mRNA of BM-derived macrophages by selecting mutant RAW264.7 cells that acquired the ability to internalize F-PRXs (Supplementary Fig. 2f). Lentiviral expression of Msr1 or Marco in mutant RAW264.7 cells restored the cells' ability to internalize not only the F-PRX family of proteins but also F-HMGB1 and F-S100A8/A9 (Fig. 2d and Supplementary Fig. 2g). Thus, MSR1 and MARCO are important scavenger receptors for the efficient internalization of major DAMPs in vitro.

The lentiviral expression of Mafb in mutant RAW264.7 cells also restored the internalization of fluorescently labeled DAMPs (Fig. 2d and Supplementary Fig. 2g). Therefore, we investigated the molecular mechanisms underlying the MAFB-enhanced internalization of DAMPs. We found that Mafb overexpression induced the expression of MSR1, but not MARCO, in mutant RAW264.7 cells and parental RAW264.7 cells (Fig. 2e). Mafb expression directly activated luciferase reporter genes driven by the Msr1 promoter in HEK293 cells (Fig. 2f). Two putative MAF-recognition elements (MARE1 and MARE2) were found in the proximal region of the Msr1 promoter. Luciferase reporter and chromatin immunoprecipitation (ChIP) assays revealed that MARE2 is a MAFB-responsive site (Fig. 2f,g). ChIP using BM-derived macrophages confirmed the binding of MAFB to the MARE2 site, but not to the MARE1 site, in the Msr1 promoter (Supplementary Fig. 3). These results suggest that MAFB directly enhances MSR1 expression in myeloid cells, thereby inducing efficient internalization of DAMPs.

MAFB enhances MSR1 expression in the delayed phase of ischemic stroke

Next, we investigated the expression levels of MSR1 and MARCO in cells collected from ischemic brain 3 d after MCAO. MSR1 and MARCO expression levels were most prominent in the CD45hiCD11bhi cell population (Fig. 3a). Cell surface MSR1 expression levels on Gr-1lo mononuclear phagocytes in the CD45hiCD11bhi population were higher than on Gr-1hi neutrophils (Supplementary Fig. 4). Results from head-shielded BM-chimeric mice confirmed that most of the MSR1hi cells in the CD45hiCD11bhi population were derived from infiltrating myeloid cells (Supplementary Fig. 1b). These results are consistent with our observations that DAMPs were efficiently internalized by infiltrating mononuclear phagocytes (Fig. 1e).

Figure 3: Mafb-dependent enhanced MSR1 expression and characteristics of the MSR1hi myeloid fraction in ischemic brain.
figure3

(a) Expression levels of MSR1 and MARCO (arbitrary units) in cells collected from day 3 post-ischemic brain (n = 5 mice per group). (b) Time-dependent changes in MSR1 and MARCO expression patterns (arbitrary units) in CD45hiCD11bhi or CD45intCD11bint cells collected from ischemic brain at the indicated time points after stroke onset (n = 5 mice per group). (c) Comparison of F-PRX, F-HMGB1, and F-S100A8/A9 internalization by MSR1hi and MSR1lo cells in the CD45hiCD11bhi population isolated from day 3 post-ischemic brain. The gates for the MSR1hi and MSR1lo fractions were set to divide cells into three fractions on the basis of MSR1 expression (arbitrary units), as shown to the left. The results were confirmed in two independent experiments. (d) MSR1hi and MSR1lo cells in the CD45hiCD11bhi population were isolated from the day 3 post-ischemic brains of 12 mice, and the relative mRNA expression levels of inflammatory or neurotrophic mediators were evaluated by real-time PCR and compared to those from sham-operated brains. Representative data from the results, which were confirmed in two independent experiments, were shown. (e) Mafb mRNA expression levels in the pooled CD45 population and the CD45hiCD11bhi population, each isolated by FACS from 12 Mafbflox/flox and 12 Lysm-Cre; Mafbflox/flox MCAO mice on day 3. Representative data are shown from the results that were confirmed in two independent experiments. (f) MSR1 expression levels in the CD45hiCD11bhi population collected from the day 1 or day 3 post-ischemic brains of Mafbflox/flox or Lysm-Cre; Mafbflox/flox mice (n = 6 mice per group). An IgG isotype control was used to measure background fluorescence as a negative control (gray dashed line). (g) MSR1 expression levels (arbitrary units) of transferred cells (carboxyfluorescein succinimidyl ester (CFSE) labeled) and recipient cells in the CD45hiCD11bhi population on day 3 after stroke onset (n = 4 mice per group). CCR2+-enriched BM myeloid cells were transferred into MCAO mice 6 h after stroke onset. **P < 0.01, ***P < 0.001 versus the CD45 population (a), day 1 (b), Mafbflox/flox mice (f), or recipient cells (g) (a,b, one-way ANOVA with Dunnett's correction; f,g, two-sided Student's t-test). n.s., not significant. Data are shown as mean ± s.e.m. Data points (circles) overlaid upon the group means in panels (a,b,f,g) represent individual values from animal.

MSR1 expression in the CD45hiCD11bhi population was elevated on days 3 and 6 after stroke onset when compared to expression on day 1, whereas MARCO expression levels in this cell population changed only slightly over this time period (Fig. 3b). MSR1 and MARCO expression levels in the CD45hiCD11bhi population were always higher than those in the CD45intCD11bint population, which included microglia (Fig. 3b). To investigate the effects of increased MSR1 expression on DAMP internalization and inflammation in the delayed phase of ischemic stroke, we used FACS to separate MSR1hi and MSR1lo cells in the CD45hiCD11bhi population from day 3 post-ischemic brains. MSR1hi cells internalized F-PRXs, F-HMGB1, and F-S100A8/A9 in vitro more efficiently than MSR1lo cells (Fig. 3c). We also found that MSR1hi cells did not express high levels of tumor necrosis factor (TNF)-α, IL-1β, or IL-23p19 when compared with MSR1lo cells but were major sources of insulin-like growth factor (IGF)-1, an important neurotrophic factor37,38 (Fig. 3d).

Next, we examined the role of MAFB in infiltrating myeloid cells in regulating the expression of Msr1 using mice with Mafb deficiency specifically in macrophages and neutrophils (Lysm (Lyz2)-Cre; Mafbflox/flox mice). We confirmed the nearly complete deletion of Mafb in the major myeloid subpopulation within the CD45hiCD11bhi population on days 1 and 3 after MCAO, although partial deletion was observed in the CD45intCD11bint population, as previously described elsewhere13,39,40 (Fig. 3e and Supplementary Fig. 5a,b). Because MAFB has been implicated in myeloid cell differentiation and proliferation41,42, we examined the phenotypes of macrophages in Lysm-Cre; Mafbflox/flox mice. There was no significant difference in the cellular population, MSR1 expression level, proliferative response of peripheral blood or BM cells to macrophage colony-stimulating factor (M-CSF), or number of microglial cells between Mafbflox/flox and Lysm-Cre; Mafbflox/flox mice during ischemic stroke (Supplementary Fig. 5c–e). Similarly, MSR1 expression levels in the CD45hiCD11bhi population on day 1 after stroke onset were not affected by Mafb deletion (Fig. 3f). These data indicate that MAFB is not involved in either the constitution of the myeloid population, as previously reported43, or the baseline expression of MSR1. Over the period from day 1 to day 3 after MCAO, however, the increase in MSR1 expression in the CD45hiCD11bhi population was almost completely diminished in Lysm-Cre; Mafbflox/flox mice without changes in MARCO expression (Fig. 3f). Mafb and MSR1 expression levels were not changed in myeloid cells from peripheral organs by ischemic brain injury (Supplementary Fig. 6a,b), suggesting that the cerebral environment induced MSR1hi cells through MAFB.

To delete Mafb more widely in monocytes and macrophages, we crossed Mafbflox/flox mice with Csf1r-CreER mice, in which Mafb can be deleted in Csf1r-expressing cells in a tamoxifen-inducible manner44. A similar defect in MSR1 induction in the CD45hiCD11bhi population was observed in these MCAO mice with myeloid-specific deficiency for Mafb (Supplementary Fig. 7a,b).

To confirm that enhanced MSR1 expression was defective in Mafb-deficient infiltrating myeloid cells in ischemic brain, we performed experiments in which CCR2+-enriched myeloid cells were transferred into non-irradiated mice in a MCAO model (Fig. 3g and Supplementary Fig. 8a) because CCR2 has been shown to be essential for the infiltration of myeloid cells into ischemic brain34 (Supplementary Fig. 8a). Mafb and MSR1 expression levels in CCR2+ myeloid cells were higher than those in CCR2 myeloid cells, whereas there was no difference in the inflammatory response to PRXs (Supplementary Fig. 8b,c). We obtained CCR2+-enriched BM myeloid cells through depletion of CD81+ cells (Supplementary Fig. 8d). The adoptive transfer of Ly5.1+CCR2+-enriched myeloid cells into Ly5.2+ MCAO mouse models enabled us to detect a sufficient number of transferred myeloid cells in ischemic brain (Supplementary Fig. 8e). The time-dependent increase in MSR1 expression in transferred wild-type (WT) Ly5.1+ cells was similar to that seen in the recipient Ly5.2+CD45hiCD11bhi population during ischemic stroke (Supplementary Fig. 8f,g). In contrast, MSR1 expression on day 3 was retained at low levels in transferred Mafb-deficient CCR2+ cells (Fig. 3g). There was no difference in either replacement ratio or proliferation between WT and Mafb-deficient transferred cells (Supplementary Fig. 8h). Thus, MAFB is a critical regulator of the increase of MSR1 expression in infiltrating myeloid cells in the delayed phase of ischemic stroke.

MSR1 and MARCO play essential roles in the clearance of DAMPs after ischemic stroke

We next examined the effect of combined Msr1 and Marco (Msr1/Marco) deficiency on the internalization of DAMPs in vitro. Myeloid cells isolated from the ischemic brains of Msr1/Marco-deficient mice did not internalize F-PRXs efficiently (Fig. 4a). Inhibition of MSR1 or MARCO with blocking antibodies revealed that MSR1 was more critical than MARCO for F-PRX internalization in vitro (Fig. 4b). Internalization of F-HMGB1 and F-S100A8/A9 was also impaired in the Msr1/Marco-deficient CD45hiCD11bhi population, suggesting that the MSR1–MARCO system is involved in the uptake of major DAMPs by mononuclear phagocytes (Fig. 4a). However, the effect of Msr1/Marco deficiency on F-HMGB1 internalization was limited, suggesting that additional mechanisms are involved in the clearance of HMGB1.

Figure 4: Impaired clearance of DAMPs due to Msr1/Marco deficiency exacerbates the pathology of ischemic stroke.
figure4

(a) Internalization of F-PRXs, F-HMGB-1, and F-S100A8/A9 in CD45hiCD11bhi cells collected from the day 3 post-ischemic brains of WT or Msr1/Marco-deficient mice in vitro (n = 5 mice per group). (b) Inhibition of F-PRX internalization by anti-MSR1 and/or anti-MARCO antibody in the CD45hiCD11bhi population (n = 5 mice per group). Cells collected from day 3 post-ischemic brains were cultured with F-PRXs and the indicated neutralizing antibodies in vitro. (c,d) Immunohistochemistry of PRX1/2, PRX5, or PRX6 with TUNEL staining in ischemic brains on day 4 after stroke onset. (c) Extracellular release of PRXs outside the cell membrane (stained by pan-cadherin) of TUNEL+ dead brain cells in Msr1/Marco-deficient mice (confocal laser scanning microscopy). Scale bars, 10 μm. (d) The PRX+ and TUNEL+ areas in ischemic brains were measured (fluorescence microscopy). Scale bars, 1 mm. (e) Relative mRNA expression levels of inflammatory mediators in cells collected from the day 3 post-ischemic brains of WT or Msr1/Marco-deficient mice (n = 5 mice per group) in comparison to those from sham-operated brains. (f,g) Infarct volumes on days 1, 4, and 7 after stroke onset (scale bar, 1 mm) (f) and neurological (motor-deficit) scores (g) in WT (n = 10) and Msr1/Marco-deficient (n = 8) mice. (h) Infarct volumes on day 28 after stroke onset in WT and Msr1/Marco-deficient mice. Scale bar, 1 mm. (i) Time-dependent changes in neurological scores at the indicated time points after stroke onset (n = 10 for each group). BL, baseline. *P < 0.05, **P < 0.01, ***P < 0.001 versus control IgG (b) or WT mice (a,di) (b, one-way ANOVA with Dunnett's correction; a,di, two-sided Student's t-test). Data are shown as mean ± s.e.m. Data points (circles) overlaid upon the group means in panels (a,b,d,e,f,h) represent individual values from animal.

We then examined whether clearance of DAMPs in ischemic brain is dependent on Msr1 and Marco. The extracellular accumulation of PRXs reaches its peak at 24 h after stroke onset and disappears thereafter in WT mice25; however, immunostaining revealed that larger amounts of PRX-containing debris had accumulated within the day 4 post-ischemic brains of Msr1/Marco-deficient mice than in WT mice (Fig. 4c,d). To confirm the extracellular release of PRXs from dead cells, we visualized the cell membrane by immunohistochemical staining of the plasma membrane marker pan-cadherin. As shown in Figure 4c, PRX+ debris was observed outside the pan-cadherin+ region of TUNEL+ dead cells in the infarct region of Msr1/Marco-deficient mice. On the other hand, the mRNA expression levels of PRX-family proteins in brain tissue during ischemic stroke were not different between WT and Msr1/Marco-deficient mice (Supplementary Fig. 9a), suggesting that accumulation of PRX proteins was due to delayed degradation but not enhanced synthesis. The extracellular release of HMGB1 and S100A8/A9 in day 4 post-ischemic brain was hardly detected by immunohistochemistry, as previously reported17,18, whereas we found that the serum HMGB1 and S100A8/A9 levels after MCAO were elevated in Msr1/Marco-deficient mice when compared with WT mice (Supplementary Fig. 9b).

MSR1, MARCO, and MAFB in infiltrating myeloid cells have dominant roles in DAMP clearance and the pathology of ischemic stroke

We next compared the inflammatory pathologies after MCAO in WT and Msr1/Marco-deficient mice. The expression levels of TNF-α, IL-1β, and IL-23p19 in cells collected from day 3 post-ischemic brain were remarkably higher in Msr1/Marco-deficient mice (Fig. 4e and Supplementary Fig. 9c). The infarct volumes in Msr1/Marco-deficient mice were similar to those in WT mice on day 1 but were significantly greater on day 4 and day 7 after stroke onset than those in WT mice (Fig. 4f). Exacerbated neurological deficits in Msr1/Marco-deficient mice were apparent beginning on day 3 after stroke onset (Fig. 4g), indicating the pivotal role of the MSR1–MARCO system in preventing excess inflammation in the delayed phase of transient MCAO. Infarct volume on day 28 after stroke onset and long-term neurological deficits were also exacerbated in Msr1/Marco-deficient mice when compared with WT mice (Fig. 4h,i; cerebral blood flow and survival rates are shown in Supplementary Table 1).

Next, we examined BM-chimeric mice in which Msr1 and Marco were deleted from either infiltrating myeloid cells or other brain cells. Significantly larger amounts of PRX-containing debris were detected in the day 4 post-ischemic brains of Msr1/Marco-deficient BM–transferred mice than in those of WT BM–transferred mice, regardless of the genotypes of the recipient mice (Fig. 5a). The infarct volume of Msr1/Marco-deficient BM–transferred mice was significantly larger than that of WT BM–transferred mice (Fig. 5b; cerebral blood flow and survival rates are shown in Supplementary Table 2). Thus, combined Msr1/Marco deficiency in infiltrating myeloid cells, but not in resident brain cells, was associated with exaggerated pathologies after MCAO.

Figure 5: The dominant roles of MSR1–MARCO and MAFB in infiltrating myeloid cells for the pathologies of ischemic stroke.
figure5

(a,b) PRX+ and TUNEL+ areas in ischemic brains (scale bar, 500 μm) (a) and infarct volume (scale bar, 1 mm) (b) of BM-chimeric mice on day 4 after stroke onset. WT or Msr1/Marco-deficient BM cells were transferred (arrows) into irradiated WT or Msr1/Marco-deficient mice. (c) PRX+ and TUNEL+ areas in ischemic brains as measured through immunohistochemical staining on day 4 after stroke onset. Scale bar, 1 mm. (d,e) Infarct volume on day 4 and neurological scores (d) and infarct volume on day 28 (e) after stroke onset in Mafbflox/flox and Lysm-Cre; Mafbflox/flox mice. Scale bars, 1 mm. (f) Time-dependent changes in neurological scores at the indicated time points after stroke onset (n = 10 mice for each group). *P < 0.05, **P < 0.01, ***P < 0.001 versus BM-transferred WT mice (a,b) or Mafbflox/flox mice (cf) (a,b, one-way ANOVA with Dunnett's correction; cf, two-sided Student's t-test). Data are shown as mean ± s.e.m. Data points (circles) overlaid upon the group means in panels (ae) represent individual values from animal.

As MAFB is important for enhanced MSR1 expression in the delayed phase of transient MCAO, we sought to clarify the importance of MAFB in infiltrating myeloid cells using Lysm-Cre; Mafbflox/flox mice. We observed significantly larger amounts of PRX1 and PRX2 (PRX1/2)-, PRX5-, and PRX6-containing debris around TUNEL+ infarcted regions on day 4 after stroke onset in Lysm-Cre; Mafbflox/flox mice than in WT mice (Fig. 5c). Infarct volume and neurological deficits were also exaggerated in Lysm-Cre; Mafbflox/flox mice when compared with Mafbflox/flox mice (Fig. 5d). Lysm-Cre; Mafbflox/flox mice also showed greater infarct volume on day 28 after stroke onset and exacerbated long-term neurological deficits (Fig. 5e,f; cerebral blood flow and survival rates are shown in Supplementary Table 3), suggesting that MAFB-dependent enhanced MSR1 expression in infiltrating myeloid cells is linked to the pathologies of ischemic stroke.

Am80 enhances Mafb and Msr1 expression and improves the pathology of ischemic stroke

To test the therapeutic relevance of our findings, we attempted to accelerate the clearance of DAMPs by promoting MAFB expression. Recently, agonists of the retinoid X receptor (RXR) have been reported to induce Mafb mRNA expression45. Among the various agonists, we found that administration of Am80, an agonist for RAR, which forms heterodimers with RXR, significantly increased Mafb expression in CD45hiCD11bhi cells on day 3 after MCAO (Fig. 6a). Administration of Am80 just after MCAO also promoted the internalization of F-PRXs by CD45hiCD11bhi cells collected from day 3 post-ischemic mouse brains in vitro (Fig. 6b). We then investigated the MSR1 expression of CD45hiCD11bhi cells and found that MSR1 expression in WT CD45hiCD11bhi cells on day 3 after stroke onset was reinforced by Am80 administration just after MCAO (Fig. 6c). In Lysm-Cre; Mafbflox/flox mice, an increase of MSR1 expression in CD45hiCD11bhi cells was not observed on day 3 after stroke onset and Am80 did not modify MSR1 expression levels (Fig. 6c). These results indicate that Am80 enhances MAFB-mediated MSR1 induction and promotes the internalization of DAMPs.

Figure 6: Therapeutic effect of Am80 through its promotion of MAFB-dependent enhanced MSR1 expression.
figure6

(a) Relative Mafb mRNA expression levels in a pooled CD45hiCD11bhi population isolated by FACS from 12 vehicle-treated and 12 Am80-treated MCAO mice on day 3 in comparison to those in CD45 cells isolated from sham-treated mice. The results were confirmed in three independent experiments. (b) Internalization of F-PRXs by the CD45hiCD11bhi population collected from the ischemic brains of vehicle- or Am80-treated mice on day 3 after stroke onset (n = 5 mice per group). (c) MSR1 expression levels (arbitrary units) of the CD45hiCD11bhi population collected from the ischemic brains of vehicle- or Am80-treated mice on day 3 after stroke onset (n = 6 mice per group). An IgG isotype control was used to measure background fluorescence as a negative control. (d) Infarct volume and neurological score on day 4 in vehicle- and Am80-treated WT mice. Vehicle or Am80 was successively administered intravenously (i.v.) at 0, 12, 24, and 36 h. Scale bar, 1 mm. (e,f) Infarct volume on day 28 (scale bar, 1 mm) (e) and neurological scores (f) in vehicle- and Am80-treated WT mice (n = 9 for each group). Vehicle or Am80 was administered successively at 24, 36, 48, and 60 h. (g) Infarct volume on day 4 in vehicle- and Am80-treated Lysm-Cre; Mafbflox/flox mice. Scale bar, 1 mm. Vehicle or Am80 was intravenously administered successively at 0, 12, 24, and 36 h (ad,g) or successively at 24, 36, 48, and 60 h (e,f) after stroke onset. *P < 0.05, **P < 0.01, ***P < 0.001 versus vehicle-treated WT mice (bf) or vehicle-treated Lysm-Cre; Mafbflox/flox mice (g) (c, one-way ANOVA with Dunnett's correction; b,dg, two-sided Student's t-test). Data are shown as mean ± s.e.m. Data points (circles) overlaid upon the group means in panels (be,g) represent individual values from animal.

Next, we investigated the pathologies of MCAO in Am80-treated mice. A number of previous studies have demonstrated the neuroprotective and anti-inflammatory effects of Am80 in various cerebral diseases46,47,48,49, yet, to our knowledge, a therapeutic effect after MCAO has not been demonstrated. We found that Am80 administration to male mice just after MCAO improved the neurological deficits and infarct volume (Fig. 6d; cerebral blood flow and survival rates are shown in Supplementary Table 4). Furthermore, Am80 improved long-term stroke outcome even when administered 24 h after stroke onset (Fig. 6e,f). A similar therapeutic effect of Am80 against ischemic brain injury was observed in female mice (Supplementary Fig. 10a). In addition to transient cerebral ischemia, Am80 administered 24 h after stroke onset was also effective in the cerebral photothrombotic ischemia model (Supplementary Fig. 10b). Am80 has been shown to be neuroprotective by various mechanisms46,47,48,49. Therefore, we examined the therapeutic effects of Am80 in Lysm-Cre; Mafbflox/flox mice (Fig. 6g) and Msr1/Marco-deficient mice (Supplementary Fig. 10c). Am80 administration just after MCAO did not result in a very strong improvement in infarct volume and neurological deficits in Lysm-Cre; Mafbflox/flox or Msr1/Marco-deficient mice (changes in cerebral blood flow during ischemic stroke and survival rates in these experiments are shown in Supplementary Table 5). These data suggest that the therapeutic effect of Am80 on ischemic brain injury is largely dependent on the MAFB–MSR1 pathway.

Discussion

Macrophages and microglia not only function as pivotal inflammatory mediators but also play an important role in the resolution of inflammation and recovery after CNS injuries. However, identification of the myeloid cells that have beneficial effects during ischemic stroke remains controversial50,51. In this study, we demonstrated the essential role of MAFB-dependent enhancement of MSR1 expression in infiltrating mononuclear phagocytes for the efficient clearance of DAMPs to resolve inflammation and prevent the exacerbation of ischemic stroke pathologies. We have shown that infiltrating mononuclear cells convert from proinflammatory to anti-inflammatory and pro-resolving phenotypes in the brain. The factors that induce such a conversion remain to be clarified. Although IL-4 is known to induce pro-resolution macrophages, the administration of an anti-IL-4 neutralizing antibody did not alter the MSR1 expression level in CD45hiCD11bhi cells on day 3 after stroke onset (data not shown). Retinol or retinoic acids may be among these unknown molecules, given that the RAR agonist Am80 enhanced Mafb expression (Fig. 6) and retinoid signaling has been shown to be important in the adult brain52. To date, however, there have been no reports of increased levels of retinoids in the brain after ischemic stroke. Identification of cerebral factors that activate MAFB may improve understanding of the detailed mechanisms involved in the resolution of sterile inflammation.

MAFB, which is strongly expressed in monocytes and macrophages, is a pivotal transcription factor for macrophage differentiation43,53. Lysm-Cre-mediated Mafb deletion does not affect the myeloid population or self-replication in response to M-CSF or post-ischemic inflammation. In accordance with this, Mafb-deficient macrophages have been reported to exhibit intact basic macrophage function under normal and lipopolysaccharide-activated conditions in mice reconstituted with Mafb-deficient fetal liver43. Although, to our knowledge, no previous reports have described the effect of MAFB on the resolution of inflammation, our data suggest that MAFB is deeply involved in controlling sterile inflammation by promoting the clearance of DAMPs.

We showed that Am80, a RAR agonist, enhanced Mafb expression in infiltrating mononuclear phagocytes, indicating that the RXR heterodimer is the important therapeutic target that accelerates the resolution of cerebral post-ischemic inflammation. Among various agonists of RXR and its heterodimeric partners, including RAR and liver X receptor (LXR), we found that Am80 was the most useful therapeutic agent for the induction of MSR1 in infiltrating mononuclear phagocytes after ischemic stroke. Am80 has been shown to have various neuroprotective and anti-inflammatory effects in various neurological disease models46,47,48,49. As for the mechanism, we propose that Am80 promotes resolution of inflammation by enhancing MSR1 expression in infiltrating mononuclear phagocytes. Am80 could be a therapeutic drug that can be administered during the subacute phase of stroke, as the process of DAMP clearance is very active within 3 d of stroke. Am80 is thus a new therapeutic agent that could potentially widen the therapeutic timeframe for treating ischemic stroke.

Methods

Mice.

Msr1/Marco-deficient and Mafbflox/flox mice were maintained in a conventional facility at Keio University in Tokyo, Japan. All experiments were approved by the Institutional Animal Research Committee of Keio University (approval 08004). Msr1/Marco-deficient mice54,55 were kindly provided by K. Tryggvason (Duke–NUS Medical School). Lysm-Cre; Mafbflox/flox mice were kindly provided by S. Takahashi (Tsukuba University). For the generation of Mafbflox/flox mice, a loxP site was inserted into exon 1 of the Mafb gene and a neomycin cassette was inserted into the 3′ downstream sequence of the Mafb gene flanked by FRT sites. The neomycin cassette was excised by crossing with FLP recombinase mice to generate Mafbflox/+ mice. Mafbflox/+ mice were bred against C57BL/6 mice to remove FLP and were subsequently bred with Lysm-Cre mice or Csf1r-Mer-iCre-Mer mice44 (purchased from the Jackson Laboratory) to delete the coding region of the Mafb gene through Cre-mediated recombination. For tamoxifen-inducible deletion of Mafb, mice were treated with 4 mg of tamoxifen (Sigma-Aldrich) dissolved in corn oil (Sigma-Aldrich) injected intraperitoneally at two time points 48 h apart (24 h before MCAO and 24 h after MCAO). Lysm-Cre; Rosa26-tdRFP mice were generated by crossing Lysm-Cre mice with Rosa26-tdRFP reporter mice56 that carried tandem-dimer red fluorescent protein (tdRFP) within the Rosa26 locus.

Mouse model of ischemic stroke.

Male C57BL/6J mice, aged 8–14 weeks and weighing 20–30 g, were randomly selected for brain ischemia experiments. Eight to 15 mice were needed to reach sufficient statistical power. The mice were anesthetized with halothane in a mixture of 70% nitrous oxide and 30% oxygen. We generated transient cerebral ischemia (MCAO, middle cerebral artery occlusion) by the insertion of a silicon-coated monofilament (Doccol Corporation) into the common carotid artery. We only included mice that showed greater than 70% reduction in cerebral blood flow of the temporal lobe (confirmed by laser Doppler flowmetry) during MCAO. During the MCAO procedure, the head temperature was kept at 36 °C. Sixty minutes after MCAO, the inserted monofilament was withdrawn to allow reperfusion. The number of excluded mice, physiological data, changes in cerebral blood flow (CBF) during ischemia, and survival rates after MCAO are shown in Supplementary Tables 1–5.

In the experiment for the photothrombotic ischemia model, we induced cortical photothrombosis by the Rose Bengal technique57. After anesthesia with halothane, body temperature was maintained at 37 °C. Anesthetized mice were placed in a stereotaxic frame, and a laser was positioned 2 mm lateral (right hemisphere) and 1 mm posterior to the bregma. Rose Bengal solution (Sigma-Aldrich) was intravenously injected at a dose of 20 mg per kg bodyweight. The dye was activated by focal illumination with a green laser (561 nm, 20 mW; Coherent) for 4 min (2 mm in diameter). The scalp was sutured, and the mice were left to recover in a 22 °C chamber.

Neurological deficits during early post-ischemic periods were evaluated according to a previously described four-point-scale method (0, no observable deficit; 1, forelimb flexion; 2, decreased resistance to lateral push without circling; 3, same behavior as grade 2 with circling) and were assessed in a blinded fashion58. For evaluation of long-term neurological function, the corner test and cylinder test were performed on days 3, 7, 14, 21, and 28 after stroke onset. Scores from these tests measured before the induction of brain ischemia are shown as baseline. For the corner test, the mouse was put into the testing apparatus, which consisted of two connected board walls forming a 30° angle. The mouse was tested for the side chosen to leave the corner, as described elsewhere59. The neurological score was calculated as (number of ipsilateral turns)/10. For the cylinder test, the mouse was placed inside a transparent cylinder and the number of independent wall placements observed for forelimb use during vertical exploration was recorded as described elsewhere59. Neurological score was calculated as (number of nonimpaired forelimb contact − number of impaired forelimb contact)/(number of nonimpaired forelimb contact + number of impaired forelimb contact + number of both types of contact).

For the evaluation of ischemic brain tissue and infarct volume, mice were sedated under deep anesthesia and transcardially perfused with PBS followed by 4% paraformaldehyde. The forebrain was removed, and 1-mm-thick coronal slices were embedded in paraffin.

The infarct area was detected by MAP2 staining and measured using ImageJ software (National Institutes of Health) in a blinded manner. Neuronal cell death in infarct regions was detected using an in situ Cell Death Detection Kit (Roche). For the immunohistochemical detection of PRX1/2-, PRX5-, and PRX6-containing debris, we used anti-PRX1/2, anti-PRX5, and anti-PRX6 antibodies, respectively; each was obtained in house from a rabbit immunized with the respective recombinant murine PRX. PRX+ areas were identified as PRX+TUNEL+ regions and measured using BZ-X700 (Keyence). The extracellular release of PRXs was examined using confocal laser scanning microscopy (LSM-710, Carl Zeiss). The immunohistochemical detection of pan-cadherin (ab16505, Abcam) in ischemic brain was performed with TUNEL staining and immunohistochemical staining of PRX1/2, PRX5, or PRX6. A detailed procedure for the immunohistochemistry has been described elsewhere25,30.

For the measurement of serum levels of HMGB1 and S100A8/A9, serum was obtained from MCAO mice on day 1 or 3 after stroke onset. Serum levels of HMGB1 from the mouse model of ischemic stroke were measured using the HMGB1 ELISA kit II (Shino-Test Corporation) following the manufacturer's protocol. This kit ensured that serum levels of HMGB1 from the mouse could be measured without interference from >80% homologous HMGB2. Serum levels of S100A8/A9 from mice after ischemic stroke were measured by the Mouse S100A8/A9 Heterodimer Duoset ELISA kit (R&D) following the manufacturer's protocol.

Am80 administration.

Precisely 1 mg of Am80 (tamibarotene; Sigma-Aldrich) was dissolved in Captisol (70 mg/ml; Chemscene) and PBS by sonication and warming to 55 °C. We confirmed that the activity of Am80 dissolved in Captisol was similar to that of Am80 in DMSO using a primary culture of BM-derived macrophages. Am80 dissolved in Captisol was intravenously administered to MCAO mice every 12 h (at 0, 12, 24, and 36 h or 24, 36, 48, and 60 h) after stroke onset.

Evaluation of DAMP internalization.

HMGB1 was purchased from Chondrex. Recombinant GST, PRX, and S100A8/A9 proteins were generated as previously described25. Briefly, cDNAs that were cloned from a mouse brain cDNA library were inserted into the pGEX6P-3 plasmid (GE Healthcare) and expressed as GST fusion proteins in BL21 competent cells (Stratagene). Following fusion protein purification using Glutathione-Sepharose 4B beads (GE Healthcare), a 100-μl slurry of protein-bound glutathione beads was washed five times with 10 ml of cold PBS. Washed, protein-bound glutathione beads were incubated with PreScission Protease (GE Healthcare) overnight at 4 °C to remove the GST tag. After incubation, the solution of recombinant protein was removed and incubated with Affi-Prep Polymyxin Support (Bio-Rad) for 12 h at 4 °C to remove endotoxins and endotoxin-bound proteins. We confirmed the purity of the recombinant proteins by means of SDS–PAGE with CBB staining (shown in Supplementary Fig. 1a) and the application of BM-derived macrophage stimulation experiments as previously described25. We also confirmed that recombinant GST protein showed no cytokine induction activity in BM-derived macrophages.

These proteins were conjugated with DyLight 488 or DyLight 650 NHS ester (Life Technologies) according to the manufacturer's instructions. After the unreactive dye was depleted by gel filtration, fluorescence-conjugated proteins were used for assays. Cells collected from day 3 post-ischemic brain or RAW264.7 cells (obtained from American Type Culture Collection: ATCC TIB-71) were incubated with fluorescence-conjugated proteins (0.3 μM) for 1–6 h at 37 °C. After several washings with RPMI-1640, cells were detached with Cell Dissociation Solution (Sigma-Aldrich) and analyzed by fluorescence microscopy or FACS. For imaging by fluorescence microscopy, LysoTracker Red DND-99 (Life Technologies) was added to the culture medium 60 min before observation by microscopy. After internalization of F-PRXs, F-HMGB1, or F-S100A8/A9, the cells collected from day 3 post-ischemic brain were incubated with rat anti–mouse F4/80 antibody (MCA497, Bio-Rad, 1:1,000) for 20 min at 37 °C. After washing with RPMI-1640, cells were incubated with Alexa Fluor 546 goat anti-rat IgG (H+L) (Invitrogen) and Hoechst 33342 for 20 min at 37 °C. After washing with RPMI-1640, cells were observed by fluorescence microscopy (BZ-X700, Keyence). For analysis by FACS, cells collected from ischemic brain or RAW264.7 cells were cultured in RPMI-1640 with 10% FBS and DyLight 650–conjugated PRX, HMGB1, or S100A8/A9 protein (0.3 μM) for 1 h. After several washings with RPMI-1640, cells were detached with Cell Dissociation Solution (Sigma-Aldrich). Detached cells from mouse ischemic brain were stained with FITC-conjugated CD45 antibody (30-F11, eBioscience, 1:400) and PerCP-conjugated CD11b antibody (M1-70, eBioscience, 1:400) and analyzed by FACS.

Random ENU mutagenesis.

We used a previously reported experimental method as a reference35. RAW264.7 cells were treated with ENU (1–3 mg/ml; Sigma-Aldrich) for 1 h at room temperature. After several washings with RPMI-1640, cells were allowed to proliferate for 48 h. ENU-treated RAW264.7 cells were incubated with DyLight 488–conjugated PRX5 (0.5 μM; used as representative of PRXs) for 1 h at 37 °C. After several washings with RPMI-1640, the cells that had not internalized PRX5 were isolated by FACSAria IIu (Becton Dickinson). For the microarray analysis of mutant RAW264.7 cells, we had Toray Industries perform the experimental procedure using 3D-Gene (Toray Industries), and we analyzed the results.

Analysis of cells collected from the brain.

Mice were perfused with PBS transcardially, and the forebrain was removed. Brain homogenates were made by mincing the brain in RPMI-1640. Brain homogenates were digested with collagenase (1 mg/ml; Sigma-Aldrich) and DNase I (50 μg/ml; Roche) for 45 min. Digested brain homogenates were then subjected to Percoll (GE Healthcare) gradient centrifugation. Cells were isolated from the interlayer between the layers corresponding to 37% Percoll and 70% Percoll. Collected cells were stained with anti-CD45 FITC (30-F11, eBioscience, 1:400), anti-CD11b PerCP-Cy5.5 (M1/70, eBioscience, 1:400), anti-MSR1 (CD204) biotin (MCA1322BT, Bio-Rad, 1:400), anti-MARCO RPE (ED31, GeneTex, 1:400), anti-Ly6G PE or APC (RB6-8C5, eBioscience, 1:1,000), anti-CD11c APC (N418, BioLegend, 1:400), and anti-F4/80 APC (BM8, eBioscience, 1:400). After several washings with PBS, cells were stained with streptavidin APC (BioLegend). Cells were washed with PBS and analyzed by FACS.

For the measurement of mRNA expression levels, RNA from collected cells was obtained using a ReliaPrep RNA Cell Miniprep System (Promega). cDNA synthesis from RNA was performed using High-Capacity cDNA Reverse Transcription Kits (Life Technologies). cDNA samples were subjected to quantitative real-time PCR analysis using SsoFast Eva Green Supermix (Bio-Rad). The ΔCT value was calculated using the CT value of hypoxanthine–guanine phosphoribosyl transferase 1 (HPRT1) converted to a relative quantification value.

Lentiviral transduction system.

cDNA was cloned into the lentivirus vector CSII-EF-MCS-IRES2-Venus (RIKEN). cDNA sequences were obtained from the CCDS database (NCBI). To generate recombinant lentivirus, cDNA expression vectors were transfected into HEK293 cells, which were obtained from the RIKEN Bioresource Center (RCB2202), along with VSV-G expression vector (pCMV-VSV-G-RSV-Rev, RIKEN) and packaging vector (pMDLg/p-RRE). Eighteen hours after transduction, the vector-containing culture medium was replaced with fresh medium, and 48 h later lentivirus-containing culture medium was collected. After concentration by centrifugation, the lentivirus pellets were resuspended in RPMI-1640 and added to the culture medium of mutant RAW267.4 cells. Thirty-six hours later, mutant RAW267.4 cells were washed with RPMI-1640 and analyzed by FACS. Venus+ cells were isolated by FACSAria IIu and were allowed to proliferate in culture medium. When we could confirm that more than 80% of the cells were Venus+, the cells were subjected to PRX internalization assays.

For the construction of a lentiviral cDNA expression library, RNA was purified from BM-derived macrophages. Oligo(dT)25-NotI-LC-LC-biotin/streptavidin Dynabeads (Life Technologies) were originally developed using an amine-reactive biotinylation agent (EZ-Link Sulfo-NHS-LC-LC-Biotin, Thermo Scientific) and amine-modified oligonucleotide. RNA from BM-derived macrophages was incubated with Oligo(dT)25 Dynabeads, and single-strand cDNA synthesis was performed with PrimeScript II (Takara). After reactions with T4 DNA polymerase (Takara) and terminal deoxynucleotidyl transferase (Takara), cDNA was amplified by PCR and cloned into the lentivirus vector CSII-EF-MCS-IRES2-Venus.

Msr1 gene disruption by the CRISPR–Cas9 system.

We used the PrecisionX Cas9 SmartNuclease system (System Biosciences) to disrupt Msr1 in RAW267.4 cells. The targeted DNA sequence was selected in exon 2 of the Msr1 gene as follows: guide RNA 1 (5′-TGAACGTGCGTCAAATTTCA-3′) and guide RNA 2 (5′-CTTCCTCACAGCACTAAAAA-3′). Per the manufacturer's instructions, we transfected Cas9 SmartNuclease vector into RAW267.4 cells by means of FuGENE 6 (Promega). Seventy-two hours later, cells were stained with anti-MSR1 antibody (MCA1322BT, Bio-Rad, 1:400) and analyzed by FACS. Msr1-deficient cells were isolated by FACSAria IIu. After performing limiting dilution, we were able to obtain Msr1-deficient clones of RAW267.4 cells (RAW ΔMSR1).

Transfer experiment with CCR2+-enriched myeloid cells.

CCR2+ (FAB5538, R&D) myeloid cells were enriched from BM and blood cells by depletion of CD3ɛ+ (145-2C11, BioLegend), B220+ (RA3-6B2, BioLegend), TER119+ (TER-119, BioLegend), CD49b+ (Dx5, BioLegend), or CD81+ (Eat-2, BioLegend) cells following positive selection with anti-CD11b antibody using magnetic-activated cell sorting. All antibodies were used at a dilution of 1:100. After sorting, the purity of CD11b+ cells (>95%) or CCR2+ cells (>40%) was assessed by FACS. CCR2+-enriched myeloid cells were prepared from Ly5.1+ mice or were labeled with CFSE before transfer. Twenty million CCR2+-enriched myeloid cells were then injected intravenously into MCAO mice at the indicated time after stroke onset. Transferred myeloid cells (Ly5.1+ or CFSE+) and recipient myeloid cells (Ly5.2+ or CFSE) that had infiltrated the ischemic brain were collected by Percoll-gradient centrifugation and analyzed by FACS.

BM chimeric mice.

Recipient mice were given lethal doses of total body irradiation (or were head-shielded) via two 5-Gy exposures given 4 h apart. Donor BM cells (1 × 107) were injected into the tail veins of irradiated recipients. Seven weeks after irradiation, these BM-chimeric mice were used in the ischemic stroke model.

Msr1 promoter assays.

The mouse Msr1 promoter was amplified by PCR from mouse genomic DNA. The amplified DNA sequences that were obtained from the Ensembl Genome Browser were cloned into pGL3-basic vectors (Promega) for promoter-driven luciferase assays. Msr1 promoter luciferase vectors, a CMV–ß-galactosidase vector (internal control), and an empty or Mafb-encoding pcDNA3.1 vector were transiently transfected into HEK293 cells using polyethyleneimine (PEI). Twenty-four hours after transfection, cells were collected for luciferase assays (Promega). In the comparison of Msr1 promoter luciferase activity between transfections with empty pcDNA3.1 vector and those with Mafb-encoding pcDNA3.1, the results were expressed as relative fold change.

Chromatin immunoprecipitation assays.

Mutant RAW264.7 cells (mock or Mafb transduced) or BM-derived macrophages (WT or Mafb deficient) were fixed with formaldehyde and then lysed for the collection of chromatin solution. The chromatin solution was sonicated using a focused ultrasonicator (Covaris S2) to obtain sheared chromatin. Control IgG or anti-Mafb antibody (Abcam, ab66506) was added to the solution of sheared chromatin at a 1:50 dilution, and MAFB-bound sheared chromatin was immunoprecipitated with magnetic Protein A beads. After the beads were washed, the immune complex was eluted and reverse cross-linking was performed with proteinase K. After DNA purification, quantitative real-time PCR was performed using the following primers: MARE1 (−517 to −417 bp) forward (5′-CATGGTCCTATAAAATGCTG-3′), MARE1 (−517 to −417 bp) reverse (5′-GTATACAGCATTTAACCAAG-3′), MARE2 (−130 to −30 bp) forward (5′-AGATTTTGCTAACTTGTGCA-3′), and MARE2 (−130 to −30 bp) reverse (5′-TAGCCCACAAGGAAAGGAAA-3′). Relative values were calculated as the ratio of immunoprecipitated DNA to input DNA.

Statistical analysis.

Data are expressed as means ± the standard error (s.e.m.). Statistical significance was determined by one-way ANOVA followed by post hoc multiple-comparisons tests (Dunnett's correction) to analyze differences among three or more groups and by unpaired Student's t-test to analyze differences between two groups. P < 0.05 was considered to represent a significant difference.

Data availability.

The microarray data are available in the Gene Expression Omnibus (GEO) database (accession number GSE89329).

Accession codes

Primary accessions

Gene Expression Omnibus

References

  1. 1

    Moskowitz, M.A., Lo, E.H. & Iadecola, C. The science of stroke: mechanisms in search of treatments. Neuron 67, 181–198 (2010).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  2. 2

    Lo, E.H. Degeneration and repair in central nervous system disease. Nat. Med. 16, 1205–1209 (2010).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  3. 3

    Chamorro, Á., Dirnagl, U., Urra, X. & Planas, A.M. Neuroprotection in acute stroke: targeting excitotoxicity, oxidative and nitrosative stress, and inflammation. Lancet Neurol. 15, 869–881 (2016).

    CAS  Article  Google Scholar 

  4. 4

    Iadecola, C. & Anrather, J. The immunology of stroke: from mechanisms to translation. Nat. Med. 17, 796–808 (2011).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  5. 5

    Zimmer, S. et al. Cyclodextrin promotes atherosclerosis regression via macrophage reprogramming. Sci. Transl. Med. 8, 333ra50 (2016).

    PubMed  PubMed Central  Article  CAS  Google Scholar 

  6. 6

    Dalli, J., Chiang, N. & Serhan, C.N. Elucidation of novel 13-series resolvins that increase with atorvastatin and clear infections. Nat. Med. 21, 1071–1075 (2015).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  7. 7

    Buckley, C.D., Gilroy, D.W., Serhan, C.N., Stockinger, B. & Tak, P.P. The resolution of inflammation. Nat. Rev. Immunol. 13, 59–66 (2013).

    CAS  Article  Google Scholar 

  8. 8

    Dirnagl, U. & Endres, M. Found in translation: preclinical stroke research predicts human pathophysiology, clinical phenotypes, and therapeutic outcomes. Stroke 45, 1510–1518 (2014).

    Article  Google Scholar 

  9. 9

    Iadecola, C. & Anrather, J. Stroke research at a crossroad: asking the brain for directions. Nat. Neurosci. 14, 1363–1368 (2011).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  10. 10

    Kono, H. & Rock, K.L. How dying cells alert the immune system to danger. Nat. Rev. Immunol. 8, 279–289 (2008).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  11. 11

    Chen, G.Y. & Nuñez, G. Sterile inflammation: sensing and reacting to damage. Nat. Rev. Immunol. 10, 826–837 (2010).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  12. 12

    David, S. & Kroner, A. Repertoire of microglial and macrophage responses after spinal cord injury. Nat. Rev. Neurosci. 12, 388–399 (2011).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  13. 13

    Prinz, M. & Priller, J. Microglia and brain macrophages in the molecular age: from origin to neuropsychiatric disease. Nat. Rev. Neurosci. 15, 300–312 (2014).

    CAS  Article  Google Scholar 

  14. 14

    Hayakawa, K., Qiu, J. & Lo, E.H. Biphasic actions of HMGB1 signaling in inflammation and recovery after stroke. Ann. NY Acad. Sci. 1207, 50–57 (2010).

    CAS  Article  Google Scholar 

  15. 15

    Qiu, J. et al. Early release of HMGB-1 from neurons after the onset of brain ischemia. J. Cereb. Blood Flow Metab. 28, 927–938 (2008).

    CAS  Article  Google Scholar 

  16. 16

    Okuma, Y. et al. Anti–high mobility group box-1 antibody therapy for traumatic brain injury. Ann. Neurol. 72, 373–384 (2012).

    CAS  Article  Google Scholar 

  17. 17

    Zhang, J. et al. Anti–high mobility group box-1 monoclonal antibody protects the blood–brain barrier from ischemia-induced disruption in rats. Stroke 42, 1420–1428 (2011).

    CAS  Article  Google Scholar 

  18. 18

    Ziegler, G. et al. Mrp-8 and -14 mediate CNS injury in focal cerebral ischemia. Biochim. Biophys. Acta 1792, 1198–1204 (2009).

    CAS  Article  Google Scholar 

  19. 19

    Vogl, T. et al. Mrp8 and Mrp14 are endogenous activators of Toll-like receptor 4, promoting lethal, endotoxin-induced shock. Nat. Med. 13, 1042–1049 (2007).

    CAS  PubMed  Article  Google Scholar 

  20. 20

    Loser, K. et al. The Toll-like receptor 4 ligands Mrp8 and Mrp14 are crucial in the development of autoreactive CD8+ T cells. Nat. Med. 16, 713–717 (2010).

    CAS  Article  Google Scholar 

  21. 21

    Klichko, V.I., Orr, W.C. & Radyuk, S.N. The role of peroxiredoxin 4 in inflammatory response and aging. Biochim. Biophys. Acta 1862, 265–273 (2016).

    CAS  Article  Google Scholar 

  22. 22

    Salzano, S. et al. Linkage of inflammation and oxidative stress via release of glutathionylated peroxiredoxin-2, which acts as a danger signal. Proc. Natl. Acad. Sci. USA 111, 12157–12162 (2014).

    CAS  Article  Google Scholar 

  23. 23

    Riddell, J.R. et al. Peroxiredoxin 1 controls prostate cancer growth through Toll-like receptor 4–dependent regulation of tumor vasculature. Cancer Res. 71, 1637–1646 (2011).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  24. 24

    Riddell, J.R., Wang, X.Y., Minderman, H. & Gollnick, S.O. Peroxiredoxin 1 stimulates secretion of proinflammatory cytokines by binding to TLR4. J. Immunol. 184, 1022–1030 (2010).

    CAS  Article  Google Scholar 

  25. 25

    Shichita, T. et al. Peroxiredoxin family proteins are key initiators of post-ischemic inflammation in the brain. Nat. Med. 18, 911–917 (2012).

    CAS  Article  Google Scholar 

  26. 26

    Kuang, X. et al. Ligustilide ameliorates neuroinflammation and brain injury in focal cerebral ischemia/reperfusion rats: involvement of inhibition of TLR4/peroxiredoxin 6 signaling. Free Radic. Biol. Med. 71, 165–175 (2014).

    CAS  Article  Google Scholar 

  27. 27

    Uzawa, A. et al. Increased serum peroxiredoxin 5 levels in myasthenia gravis. J. Neuroimmunol. 287, 16–18 (2015).

    CAS  Article  Google Scholar 

  28. 28

    Rashidian, J. et al. Essential role of cytoplasmic cdk5 and Prx2 in multiple ischemic injury models, in vivo. J. Neurosci. 29, 12497–12505 (2009).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  29. 29

    Dayon, L. et al. Brain extracellular fluid protein changes in acute stroke patients. J. Proteome Res. 10, 1043–1051 (2011).

    CAS  Article  Google Scholar 

  30. 30

    Shichita, T. et al. Pivotal role of cerebral interleukin-17-producing γδ T cells in the delayed phase of ischemic brain injury. Nat. Med. 15, 946–950 (2009).

    CAS  PubMed  Article  Google Scholar 

  31. 31

    Ito, M. et al. Bruton's tyrosine kinase is essential for NLRP3 inflammasome activation and contributes to ischaemic brain injury. Nat. Commun. 6, 7360 (2015).

    PubMed  PubMed Central  Article  Google Scholar 

  32. 32

    Benakis, C., Garcia-Bonilla, L., Iadecola, C. & Anrather, J. The role of microglia and myeloid immune cells in acute cerebral ischemia. Front. Cell. Neurosci. 8, 461 (2015).

    PubMed  PubMed Central  Article  Google Scholar 

  33. 33

    Mildner, A. et al. Microglia in the adult brain arise from Ly-6ChiCCR2+ monocytes only under defined host conditions. Nat. Neurosci. 10, 1544–1553 (2007).

    CAS  Article  Google Scholar 

  34. 34

    Gliem, M. et al. Macrophages prevent hemorrhagic infarct transformation in murine stroke models. Ann. Neurol. 71, 743–752 (2012).

    CAS  Article  Google Scholar 

  35. 35

    Schlegel, J., Neff, F. & Piontek, G. Serial induction of mutations by ethylnitrosourea in PC12 cells: a new model for a phenotypical characterization of the neurotoxic response to 6-hydroxydopamine. J. Neurosci. Methods 137, 215–220 (2004).

    CAS  Article  Google Scholar 

  36. 36

    Canton, J., Neculai, D. & Grinstein, S. Scavenger receptors in homeostasis and immunity. Nat. Rev. Immunol. 13, 621–634 (2013).

    CAS  Article  Google Scholar 

  37. 37

    Lalancette-Hébert, M., Gowing, G., Simard, A., Weng, Y.C. & Kriz, J. Selective ablation of proliferating microglial cells exacerbates ischemic injury in the brain. J. Neurosci. 27, 2596–2605 (2007).

    PubMed  PubMed Central  Article  CAS  Google Scholar 

  38. 38

    Li, S. et al. An age-related sprouting transcriptome provides molecular control of axonal sprouting after stroke. Nat. Neurosci. 13, 1496–1504 (2010).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  39. 39

    Goldmann, T. et al. A new type of microglia gene targeting shows TAK1 to be pivotal in CNS autoimmune inflammation. Nat. Neurosci. 16, 1618–1626 (2013).

    CAS  Article  Google Scholar 

  40. 40

    Wieghofer, P., Knobeloch, K.P. & Prinz, M. Genetic targeting of microglia. Glia 63, 1–22 (2015).

    Article  Google Scholar 

  41. 41

    Aziz, A., Soucie, E., Sarrazin, S. & Sieweke, M.H. MafB/c-Maf deficiency enables self-renewal of differentiated functional macrophages. Science 326, 867–871 (2009).

    CAS  Article  Google Scholar 

  42. 42

    Soucie, E.L. et al. Lineage-specific enhancers activate self-renewal genes in macrophages and embryonic stem cells. Science 351, aad5510 (2016).

    PubMed  PubMed Central  Article  Google Scholar 

  43. 43

    Aziz, A. et al. Development of macrophages with altered actin organization in the absence of MafB. Mol. Cell. Biol. 26, 6808–6818 (2006).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  44. 44

    Qian, B.Z. et al. CCL2 recruits inflammatory monocytes to facilitate breast-tumour metastasis. Nature 475, 222–225 (2011).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  45. 45

    Hamada, M. et al. MafB promotes atherosclerosis by inhibiting foam-cell apoptosis. Nat. Commun. 5, 3147 (2014).

    Article  CAS  Google Scholar 

  46. 46

    Matsushita, H. et al. A retinoic acid receptor agonist Am80 rescues neurons, attenuates inflammatory reactions, and improves behavioral recovery after intracerebral hemorrhage in mice. J. Cereb. Blood Flow Metab. 31, 222–234 (2011).

    CAS  Article  Google Scholar 

  47. 47

    Katsuki, H. et al. Retinoic acid receptor stimulation protects midbrain dopaminergic neurons from inflammatory degeneration via BDNF-mediated signaling. J. Neurochem. 110, 707–718 (2009).

    CAS  Article  Google Scholar 

  48. 48

    Klemann, C. et al. Synthetic retinoid AM80 inhibits Th17 cells and ameliorates experimental autoimmune encephalomyelitis. Am. J. Pathol. 174, 2234–2245 (2009).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  49. 49

    Takeuchi, H. et al. Retinoid X receptor agonists modulate Foxp3+ regulatory T cell and Th17 cell differentiation with differential dependence on retinoic acid receptor activation. J. Immunol. 191, 3725–3733 (2013).

    CAS  Article  Google Scholar 

  50. 50

    Desestret, V. et al. In vitro and in vivo models of cerebral ischemia show discrepancy in therapeutic effects of M2 macrophages. PLoS One 8, e67063 (2013).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  51. 51

    Hu, X. et al. Microglia/macrophage polarization dynamics reveal novel mechanism of injury expansion after focal cerebral ischemia. Stroke 43, 3063–3070 (2012).

    CAS  Article  Google Scholar 

  52. 52

    Lane, M.A. & Bailey, S.J. Role of retinoid signalling in the adult brain. Prog. Neurobiol. 75, 275–293 (2005).

    CAS  Article  Google Scholar 

  53. 53

    Sarrazin, S. et al. MafB restricts M-CSF-dependent myeloid commitment divisions of hematopoietic stem cells. Cell 138, 300–313 (2009).

    CAS  Article  Google Scholar 

  54. 54

    Suzuki, H. et al. A role for macrophage scavenger receptors in atherosclerosis and susceptibility to infection. Nature 386, 292–296 (1997).

    CAS  PubMed  Article  Google Scholar 

  55. 55

    Chen, Y. et al. Defective microarchitecture of the spleen marginal zone and impaired response to a thymus-independent type 2 antigen in mice lacking scavenger receptors MARCO and SR-A. J. Immunol. 175, 8173–8180 (2005).

    CAS  Article  Google Scholar 

  56. 56

    Muto, G. et al. TRAF6 is essential for maintenance of regulatory T cells that suppress Th2 type autoimmunity. PLoS One 8, e74639 (2013).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  57. 57

    Yao, H. et al. Photothrombotic middle cerebral artery occlusion and reperfusion laser system in spontaneously hypertensive rats. Stroke 34, 2716–2721 (2003).

    Article  Google Scholar 

  58. 58

    Bederson, J.B. et al. Rat middle cerebral artery occlusion: evaluation of the model and development of a neurologic examination. Stroke 17, 472–476 (1986).

    CAS  Article  Google Scholar 

  59. 59

    Balkaya, M., Kröber, J.M., Rex, A. & Endres, M. Assessing post-stroke behavior in mouse models of focal ischemia. J. Cereb. Blood Flow Metab. 33, 330–338 (2013).

    CAS  Article  Google Scholar 

Download references

Acknowledgements

We thank N. Shiino, A. Ino, and M. Asakawa for their technical assistance and H. Yao for his technical advice on the MCAO model. Msr1/Marco-deficient mice were kindly provided by K. Tryggvason (Duke–NUS Medical School). This work was supported by PRESTO from the Japan Science and Technology Agency (T.S.), a Grant-in-Aid for Scientific Research on Innovative Areas (Homeostatic regulation by various types of cell death) (15H01387) from the Ministry of Education, Culture, Sports, Science and Technology of Japan (MEXT) (T.S.), JSPS KAKENHI Grants-in-Aid for Young Scientists (B) (26870571) (T.S.) and (S) (25221305) (A.Y.), Advanced Research & Development Programs for Medical Innovation (AMED-CREST) (A.Y.), a Toray Science and Technology Grant (T.S.), the Takeda Science Foundation (T.S.), the Mochida Memorial Foundation for Medical and Pharmaceutical Research (T.S.), a Japan Heart Foundation Research Grant (T.S.), the SENSHIN Medical Research Foundation (A.Y.), Keio Gijuku Academic Developmental Funds (A.Y.), and Open Research for Young Academics and Specialists from MEXT (A.Y.).

Author information

Affiliations

Authors

Contributions

T.S. designed and performed the experiments, analyzed the data, and wrote the manuscript; Y.N., M.I., and K.K. performed the experiments; R.M. and H.O. provided technical advice; R.K. and S.T. provided Lysm-Cre; Mafbflox/flox mice; T.K. provided Msr1/Marco double-knockout mice; and A.Y. directed the entire study, designed the experiments, and wrote the manuscript.

Corresponding authors

Correspondence to Takashi Shichita or Akihiko Yoshimura.

Ethics declarations

Competing interests

The authors declare no competing financial interests.

Supplementary information

Supplementary Text and Figures

Supplementary Figures 1–10 and Supplementary Tables 1–5 (PDF 5614 kb)

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Shichita, T., Ito, M., Morita, R. et al. MAFB prevents excess inflammation after ischemic stroke by accelerating clearance of damage signals through MSR1. Nat Med 23, 723–732 (2017). https://doi.org/10.1038/nm.4312

Download citation

Further reading

Search

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing