Resolvin D2-G-Protein Coupled Receptor 18 Enhances Bone Marrow Function and Limits Steatosis and Hepatic Collagen Accumulation in Aging

Abstract

Aging is associated with nonresolving inflammation and tissue dysfunction. Resolvin D2 (RvD2) is a proresolving ligand that acts through the G-protein-coupled receptor called GPR18. Unbiased RNA sequencing revealed increased Gpr18 expression in macrophages from old mice, and in livers from elderly humans, which was associated with increased steatosis and fibrosis in middle-aged (MA) and old mice. MA mice that lacked GPR18 on myeloid cells had exacerbated steatosis and hepatic fibrosis, which was associated with a decline in Mac2⁺ macrophages. Treatment of MA mice with RvD2 reduced steatosis and decreased hepatic fibrosis, correlating with increased Mac2⁺ macrophages, increased monocyte-derived macrophages, and elevated numbers of monocytes in the liver, blood, and bone marrow. RvD2 acted directly on the bone marrow to increase monocyte-macrophage progenitors. A transplantation assay further demonstrated that bone marrow from old mice facilitated hepatic collagen accumulation in young mice. Transient RvD2 treatment to mice transplanted with bone marrow from old mice prevented hepatic collagen accumulation. Together, this study demonstrates that RvD2-GPR18 signaling controls steatosis and fibrosis and provides a mechanistic-based therapy for promoting liver repair in aging.

Graphical abstract

Figure 1

Differentially expressed genes and pathways in zymosan-elicited peritoneal macrophages from young (Y) and old (O) mice. RNA sequencing was performed on zymosan-elicited peritoneal macrophages (MΦ) from young (aged 2 months) or old (aged 18 months) mice, as described in Materials and Methods. A: Principal component (PC) analysis for the young or aged mice is shown. B and C: Normalized counts for Gpr18 expression in zymosan-elicited macrophages (B) and murine bone marrow macrophages (C) are shown. D: iPathway Guide analysis shows biological pathways with significant perturbation accummulation (pAcc) and overrepresentation (pORA) in old compared with young mice. Significant pathways are shown in red, and nonsignificant pathways are shown in black. E: Normalized counts of GPR18 expression in human liver are shown. Each dot represents an individual mouse or human. Results are means ± SEM (B, C, and E). ∗P ≤ 0.05 (t-test). NAFLD, nonalcoholic fatty liver disease; TGF-β, transforming growth factor-β.

Figure 2

Middle-aged (MA) and old (O) mice have elevated steatosis and hepatic fibrosis compared with young (Y) mice. Livers from young (aged 2 months), middle-aged (aged 11 months), or old (aged 18 months) mice were harvested. A: Frozen liver sections were stained with oil red O, and lipid deposits are visualized as red. B: Quantification of oil red O was done by ImageJ software version 1.53q analysis. C: Liver sections were stained with hematoxylin and eosin (H&E), and representative images are shown. D: Liver homogenates were assessed for triglyceride (TG) content. E and F: Representative images and quantification of picrosirius red as percentage of red area per ×4 visual field. A–C and E: Images were acquired on an Olympus microscope, and the quantification was done by ImageJ software analysis. G–J: Representative images and quantification of liver sections that were immunostained with anti–collagen IV (Col-IV) antibody (red) or anti-Mac2 antibody (green; I). G and I: Images were acquired on a Leica confocal microscope and quantification of Col-IV mean fluorescence intensity (MFI; H) and the number of Mac2⁺ cells/×40 visual field were enumerated with ImageJ software (J). Each dot represents an individual mouse. Results are means ± SEM (B, D, F, H, and J). ∗P < 0.05 (Kruskal-Wallis and Dunn post-test). Scale bars: 20 μm (A); 100 μm (C and E); 75 μm (G and I).

Figure 3

Middle-aged (MA) Gpr18 myeloid knockout (mKO) mice have increased liver triglyceride, collagen, and α-smooth muscle cells (α-SMCs) and decreased macrophages compared with middle-aged fl/fl mice. Livers from young (Y; aged 4 months) and MA (aged 12 to 14 months) Grp18 fl/fl or mKO mice were evaluated. A: Frozen liver sections were stained with oil red O, and lipid deposits are visualized as red. Oil red O was quantified as in Figure 2. B–D: Paraffin-embedded livers from fl/fl and mKO mice were immunostained for Col1a1 (B), α-SMC (C), and Mac2 (D). Images were acquired on a Leica confocal microscope, and quantification of Col1a1 mean fluorescence intensity (MFI), α-SMC, and Mac2⁺ positive cells was done with ImageJ software version 1.53q. Each dot represents an individual mouse. Results are means ± SEM (A–D). ∗P < 0.05 (one-way analysis of variance with the Tukey multiple-comparison test). Scale bars: 20 μm (A); 75 μm (B–D).

Figure 4

Resolvin D2 (RvD2) limits steatosis, reduces hepatic fibrosis, and increases Mac2⁺ macrophages and monocytes in the livers from middle-aged (MA) mice. MA (aged 11 months) mice were treated with vehicle (Veh) or RvD2 (250 ng per mouse, intraperitoneally) for 7 days. Livers were harvested as in Figure 2. A: Liver homogenates were assessed for triglyceride (TG) content. B: Picrosirious red was analyzed by ImageJ software version 1.53q analysis and represented as fold change of Veh-treated mice. C–E: Collagen IV (C), α-smooth muscle cell (α-SMC; D), and Mac2 (E) acquired on a Leica confocal microscope, and quantification was done with ImageJ software. F–H: Liver was processed and analyzed by flow cytometry for Kupffer cells (KCs; Ly6G⁻, CD11b^lo, F4/80^hi; F), monocyte-derived macrophages (mono-macs; Ly6G⁻, CD11b⁺, F4/80⁺; G) and monocytes (Ly6G⁻, CD11b⁺, Ly6C^hi; H). I: Circulating monocytes were quantified by a Heska-HT5 analyzer. Each dot represents an individual mouse. Results are mean ± SEM (A–I). ∗P < 0.05, ∗∗P < 0.01 (U-test).

Figure 5

Resolvin D2 (RvD2) increases bone marrow monocytes and monocyte progenitors. Bone marrow was harvested, and monocytes were enumerated by flow cytometry. A and B: Monocyte (CD11b⁺, Ly6G⁻, Ly6C^hi) frequencies and numbers in middle-aged (MA) mice treated with vehicle (Veh) or RvD2 are shown. C: Frequencies of phenotypic hematopoietic stem and progenitor cells in bone marrow are shown as the percentage of each population among total lineage-negative, c-Kit⁺, and Sca-1⁺ cells (LSK). D and E: The absolute frequencies and number of granulocyte-macrophage multipotent progenitors (MPP_GM) are shown. F and G: Bone marrow was obtained from MA mice and incubated for 30 minutes with vehicle or RvD2 (1 nmol/L) and plated in Methocult media. After 7 days, colonies were counted. G: Colony-forming units (CFUs) are shown for macrophages (CFU-M), granulocytes (CFU-G), granulocyte/macrophage (CFU-GM), and granulocyte/erythrocyte/monocyte/megakaryocyte (CFU-GEMM) from MA bone marrow treated with Veh or RvD2. Results are from bone marrow. Each dot represents an individual mouse. Results are means ± SEM (A–E and G). n = 4 separate mice per treatment (G). ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001 (U-test). HSC, hematopoietic stem cell. MPP_Ly, lymphocyte MPP; MPP_Mk/E, megakaryocyte/erythrocyte MPP.

Figure 6

Resolvin D2 (RvD2) limits age-related hepatic collagen accumulation in part via changes in the bone marrow. A: Schematic of competitive bone marrow transplants where young mice received either young (Y) or old (O) donor bone marrow with or without RvD2 treatment. Recipient mice received injections of either vehicle (Veh) or RvD2 for 1 week, and mice were analyzed 4 months after transplant. B: Representative images and quantification of liver sections stained with anti–collagen IV (Col-IV) antibody shown in red. C and D: Images were acquired on a Leica confocal microscope, and quantification of Col-IV (C) or Col1a1 (D) mean fluorescence intensity (MFI) was done with ImageJ software version 1.53q. Each dot represents an individual mouse. Results are means ± SEM (C and D). ∗P < 0.05 (one-way analysis of variance with Kruskal-Wallis and a Dunn post-test). Scale bar = 75 μm (B). WBM, whole bone marrow.

Supplemental Figure S1

Macrophages from aged mice have an increased cyclooxygenase-2 (COX2) phosphorylated p38 (p-p38) and gamma histone H2AX (γH2AX), which is decreased by resolvin D2 (RvD2). A: Gene Set Enrichment Analysis from the RNA sequencing performed in Figure 1 is shown. B–D: Peritoneal macrophages from young (Y) or old (O) mice treated with RvD2 (described in Materials and Methods) were isolated, cultured overnight, and immunostained for COX-2 (B), p-p38 (C), or γH2AX (D) and are shown in magenta, whereas nuclei are stained in blue with DAPI. Each symbol represents an individual mouse. Results are means ± SEM (B–D). ∗P < 0.05, ∗∗P < 0.01, and ∗∗∗∗P < 0.0001 (one-way analysis of variance with the Tukey multiple-comparison test). Scale bars = 50 μm (B–D). Original magnification, ×40 (B–D). MAPK, mitogen-activated protein kinase; MFI, mean fluorescence intensity; ROS, reactive oxygen species; Veh, vehicle.

Supplemental Figure S2

Livers from middle-aged individuals have increased fibrosis. Autopsy cases containing sampled liver, without known premortem liver disease and without a liver-related cause of death, were assessed for fibrosis according to Materials and Methods. Percentage young age cohort (aged 3 to 32 years; top panel) and percentage middle-aged cohort (aged 35 to 54 years; bottom panel) who had Non-Alcoholic Steatohepatitis Clinical Research Network (NASH-CRN) scores 0 to 4. The scores are as follows: no fibrosis (score 0), perisinusoidal or portal fibrosis (score 1), perisinusoidal and portal fibrosis (score 2), bridging fibrosis (score 3), and cirrhosis (score 4).

Supplemental Figure S3

Generation of humanized GPR18 floxed mice. A: Scheme of humanized Gpr18 floxed and Gpr18-Lyzm [myeloid knockout (mKO)] mice was made as described in Materials and Methods. B and C: Bone marrow–derived macrophages from c57Bl7/6, fl/fl, and mKO mice were subjected to real-time quantitative PCR and assessed for levels of murine Gpr18 (B) and human GPR18 (C). D: Phagocytosis was performed as described in Materials and Methods, and data were assessed by fluorescence plate reader and represented as percentage increase of phagocytosis by resolvin D2 (RvD2). Each symbol represents macrophages from an individual mouse. Results are means ± SEM (B–D). ∗P < 0.05 (one-way analysis of variance with the Tukey multiple-comparison test).

Supplemental Figure S4

Resolvin D2 (RvD2) treatment to middle-aged (MA) mice does not change body weight and insulin but does decrease circulating C-reactive protein (CRP) levels. A and B: MA mice were treated with vehicle (Veh) or RvD2, as described in Materials and Methods, and body weights (A) and liver/body weights (B) were calculated. C: Serum was collected, and insulin was measured by enzyme-linked immunosorbent assay (ELISA) analysis. D–F: Plasma was collected and subjected to ELISA analysis for circulating levels of triglycerides (TGs; D), collagen IV (Col-IV; E) and CRP (F). Each symbol represents an individual mouse. Results are means ± SEM (A–F). ∗P < 0.05 (t-test). AU, arbitrary unit.

Supplemental Figure S5

Resolvin D2 (RvD2) treatment to middle-aged mice regulates genes associated with lipid metabolism. Livers were harvested and subjected to real-time quantitative PCR. Lipid metabolism genes, including Acaca (A), Scd (B), and Fas (C), were analyzed. Each symbol represents an individual mouse. Results are means ± SEM (A–C). ∗P < 0.05 (t-test). Veh, vehicle.

Supplemental Figure S6

Representative images of liver sections from middle-aged vehicle (Veh)– or resolvin D2 (RvD2)–treated mice. Experiments were performed as in Figure 4. The representative images from picrosirious (A), collagen IV (Col-IV; B), α-smooth muscle cell actin (α-SMCa; C), and Mac2 (D) are shown. Scale bars: 200 μm (A); 75 μm (B–D).

Supplemental Figure S7

Resolvin D2 (RvD2) treatment to middle-aged mice regulates several genes associated with inflammation and tissue repair. Livers were harvested and subjected to real-time quantitative PCR. A–D:Acta2 (A), Il1b (B), Mmp2 (C), and Mmp9 (D) expression levels were analyzed, and mRNA log fold change was calculated. RNAscope was performed, as described in Materials and Methods. Briefly, probes for adhesion G-protein coupled Rrceptor-E1 (Adgre1) (ie, F4/80) and metalloproteinase-9 (Mmp9) were used, and slides were counterstained with DAPI. E: The number of Mmp9 puncta per Adgre1-expressing cell was quantified and presented as average number of Mmp9 puncta per macrophage. F: The percentage of macrophages was enumerated by quantifying the number of Adgre1⁺ cells per DAPI cells per ×40 visual field. Each symbol represents an individual mouse. Results are means ± SEM (A–F). ∗P < 0.05 (t-test). Veh, vehicle.

Supplemental Figure S8

Gating strategy for liver myeloid cells. Livers were processed for flow cytometry and stained with antibodies against myeloid markers, followed by analysis. Kupffer cells (KCs) were identified as CD45.2⁺, Ly6G⁻, CD11b^lo/−, and F4/80^hi; monocytes-macrophages (MonoMac) were identified as CD45.2⁺, Ly6G⁻, CD11b⁺, Ly6C^lo/−, and F4/80⁺; and monocytes (Monos) were identified as CD45.2⁺, Ly6G⁻, CD11b⁺, and Ly6C^hi. Cells were analyzed on a BD FACSymphony or a Cytek Northern Lights instrument. FSC, forward scatter.

Supplemental Figure S9

Gating strategy for bone marrow hematopoietic stem and progenitor cells (HSPCs). Bone marrow was isolated from the hind limbs of middle-aged (MA) + vehicle (Veh) and MA + resolvin D2 (RvD2) mice. Single cells were first gated on lack of lineage markers (CD45R, Gr-1, Ter-119, CD11b, and CD3). HSPCs were identified as Lin⁻, cKit⁺, Sca-1⁺. multipotent progenitor (MPP)_Ly cells were identified as CD150⁻, CD135⁺. Hematopoietic stem cells (HSCs) were CD135⁻, CD150⁺, CD48⁻; MPP_Mk/E cells were CD135⁻, CD150⁺, CD48⁺; MPP_GM cells were CD135⁻, CD150⁻, CD48⁺; MPPs were CD135⁻, CD150⁻, CD48⁻. Cells were analyzed on a BD FACSymphony or Cytek Northern Lights instrument. FSC, forward scatter.