Inhibition of atherogenesis by the COP9 signalosome subunit 5 in vivo

Abstract

Constitutive photomorphogenesis 9 (COP9) signalosome 5 (CSN5), an isopeptidase that removes neural precursor cell-expressed, developmentally down-regulated 8 (NEDD8) moieties from cullins (thus termed "deNEDDylase") and a subunit of the cullin-RING E3 ligase-regulating COP9 signalosome complex, attenuates proinflammatory NF-κB signaling. We previously showed that CSN5 is up-regulated in human atherosclerotic arteries. Here, we investigated the role of CSN5 in atherogenesis in vivo by using mice with myeloid-specific Csn5 deletion. Genetic deletion of Csn5 in Apoe^-/- mice markedly exacerbated atherosclerotic lesion formation. This was broadly observed in aortic root, arch, and total aorta of male mice, whereas the effect was less pronounced and site-specific in females. Mechanistically, Csn5 KO potentiated NF-κB signaling and proinflammatory cytokine expression in macrophages, whereas HIF-1α levels were reduced. Inversely, inhibition of NEDDylation by MLN4924 blocked proinflammatory gene expression and NF-κB activation while enhancing HIF-1α levels and the expression of M2 marker Arginase 1 in inflammatory-elicited macrophages. MLN4924 further attenuated the expression of chemokines and adhesion molecules in endothelial cells and reduced NF-κB activation and monocyte arrest on activated endothelium in vitro. In vivo, MLN4924 reduced LPS-induced inflammation, favored an antiinflammatory macrophage phenotype, and decreased the progression of early atherosclerotic lesions in mice. On the contrary, MLN4924 treatment increased neutrophil and monocyte counts in blood and had no net effect on the progression of more advanced lesions. Our data show that CSN5 is atheroprotective. We conclude that MLN4924 may be useful in preventing early atherogenesis, whereas selectively promoting CSN5-mediated deNEDDylation may be beneficial in all stages of atherosclerosis.

Keywords: COP9 signalosome; MLN4924; NEDDylation; atherosclerosis; inflammation.

Fig. 1.

Myeloid-specific deletion of Csn5 increases atherosclerosis. Male Csn5^wt/Apoe^−/− and Csn5^Δmyeloid/Apoe^−/− mice consumed a high-fat diet for 12 wk. Lesion size in aortic root (A) and aorta (B), as well as aortic arch (C), was quantified. Data are means ± SD of n ≥ 8 mice and representative stainings. Two-tailed t test was performed for comparison of Csn5^wt/Apoe^−/− vs. Csn5^Δmyeloid/Apoe^−/−; *P < 0.05).

Fig. 2.

Csn5 deficiency promotes a proinflammatory profile in macrophages. BMDMs were isolated from Csn5^wt/Apoe^−/− and Csn5^Δmyeloid/Apoe^−/− mice. (A) Cells were stimulated with LPS (100 ng/mL) for 6 and 24 h or left untreated as indicated. Proinflammatory gene expression was determined at baseline (Left), after 6 h (Middle), and after 24 h (Right), and normalized to Gapdh. (B) Secretion of Il-12p70 was measured in supernatants of cells stimulated with LPS (100 ng/mL, 24 h; Left) or TNF-α (20 ng/mL, 24 h; Right). (C) Immunodetection of IκB-α and α-tubulin in cell lysates of BMDMs treated with LPS for indicated time points: (Top) quantification and (Bottom) representative blot. (D) Cells were stimulated with TNF-α (20 ng/mL; 0–4 h), and p65 activity was quantified relative to untreated control. (E) After stimulation with LPS (100 ng/mL), miR-146a, miR-21, and miR-155a were quantified, normalized to U6, and expressed relative to unstimulated control. (F) Secretion of MIF was measured in supernatants of BMDMs from Csn5^wt/Apoe^−/− vs. Csn5^Δmyeloid/Apoe^−/− mice following culture for 24 h. (G) HIF-1α was quantified in nuclear lysates from LPS-stimulated (100 ng/mL, 6 h) cells or untreated control cells. HIF-1α levels were normalized to lamin B1. Representative Western blot is shown. Graphs represent means ± SD of three to six independent experiments, except E and F (n ≥ 6). Mann–Whitney test (F and G) and two-way ANOVA with Bonferroni posttest were performed for comparison between Csn5^wt/Apoe^−/− and Csn5^Δmyeloid/Apoe^−/− (asterisks) and stimulated vs. nonstimulated conditions of same genotype (§; B–E; */^§P < 0.05; **/^§§P < 0.01; ***/^§§§P < 0.001).

Fig. 3.

MLN4924 inhibits proinflammatory cytokine expression and skews macrophage polarization toward M2: involvement of NF-κB, HIF-1α, and ERK signaling. Apoe^−/− BMDMs were exposed to MLN4924 or DMSO control solution and treated with LPS (100 ng/mL) or left untreated. (A) Inhibition of Cullin 1-NEDD8 by MLN4924. (B) Proinflammatory gene expression at baseline (Left) and 6 h (Middle) or 24 h (Right) of LPS stimulation, quantified by RT-PCR (normalized to Gapdh). (C) Effect of MLN4924 or DMSO control on LPS-induced secretion of TNF-α and IL-10 from Apoe^−/− BMDMs as measured by ELISA. (D) Inhibition of LPS (100 ng/mL)-induced p65 DNA binding activity by MLN4924; quantification relative to DMSO control and untreated. (E) MLN4924 stabilizes nuclear HIF-1α levels in LPS-treated BMDMs. YY1 was used as nuclear loading control. (F) Inhibition of LPS (100 ng/mL)-triggered ERK1/2 MAPK phosphorylation by MLN4924 pretreatment (Left, quantification; Right, representative Western blot); time course from 0 to 120 min relative to DMSO control and untreated. Total ERK1/2 and tubulin were used as loading control. (G) MLN4924 induces gene expression of M2 markers in LPS-stimulated BMDMs. Baseline (Left), 6 h LPS (Middle), and 24 h LPS (Right) normalized to Gapdh. (B–G) Graphs represent means ± SD of n = 7 (F) or n = 3 (B–E and G) independent experiments. Two-way ANOVA with Bonferroni posttest was performed for comparison of DMSO vs. MLN4924-treated cells (asterisks) and stimulated vs. nonstimulated conditions within DMSO- or MLN-treated group (§; */^§P < 0.05; **/^§§P < 0.01; ***/^§§§P < 0.001).

Fig. 4.

MLN4924 attenuates proatherosclerotic gene expression, NF-κB activation, and monocyte arrest in inflammatory-elicited endothelial cells while stabilizing HIF-1α. HUVECs were pretreated with MLN4924 or DMSO control solvent and challenged with TNF-α (20 ng/mL) or LPS (100 ng/mL) as indicated. (A) Real-time RT-PCR analysis of CCL2, ICAM-1, VCAM-1, or P-selectin mRNA levels in TNF-α–stimulated HUVECs. mRNA levels were normalized to GAPDH and expressed relative to unstimulated controls. Data represent means ± SD of one representative experiment among four, each performed in triplicate. (B) MLN4924 suppresses monocyte adhesion on inflammatory endothelium. Quantification of Calcein-AM–labeled THP1 cells adhered to TNF-α–activated HUVEC monolayers pretreated with MLN4924 or DMSO control. Data represent means ± SD of four independent experiments. (C and D) MLN4924 attenuates NF-κB activity in endothelial cells. (C) Immunodetection of CUL1, p-IκB-α, and α-tubulin in cell lysates from HUVECs treated as indicated. Immunoblots are representative of three independent experiments. (D) NF-κB–driven luciferase activity in HUVECs stimulated with TNF-α (n = 6) after normalization to β-gal relative to untreated control. (E) MLN4924 stabilizes HIF-1α in endothelium. Quantification of HIF-1α levels in nuclear lysates of cells stimulated with LPS vs. untreated. HIF-1α levels were normalized to lamin B1 (n = 4). Representative blot is shown. (A–E) Data represent means ± SD. Two-way ANOVA with Bonferroni posttest was performed for comparison of MLN4924- vs. DMSO control solution-treated cells (asterisks) and stimulated vs. nonstimulated conditions within DMSO- or MLN-treated group (§; */^§P < 0.05, **/^§§P < 0.01, and ***/^§§§P < 0.001).

Fig. 5.

MLN4924 treatment in vivo reduces the size of early atherosclerotic lesions in aorta and aortic root. Male Apoe^−/− mice consumed a high-fat diet for 12 wk; in the middle of the diet, mice were treated with 60 mg/kg MLN4924 (12w-MLN4924) or DMSO control solution (12w-DMSO) for 6 wk. Mice that consumed a diet for 6 wk (6w) without any further treatment were used as proof-of-model control. (A) Absolute numbers of peripheral blood cells. (B) MLN4924 reduces atherosclerotic lesions in aorta. Quantification of lesions in aorta with representative oil red O stainings shown. (C) MLN4924 reduces early but not advanced atherosclerotic lesions in aortic root. Quantification of lesions in total aortic root (Left) and after classification into early/intermediate (Middle) vs. advanced lesions (Right). Shown are representative oil red O stainings. (A–C) Data represent means ± SD of n ≥ 6 mice. Two-tailed t test or Mann–Whitney test was performed as appropriate for comparison between MLN- vs. DMSO-treated groups (A); one-way ANOVA with Newman–Keuls or Dunn’s posttest was performed as appropriate for comparison of 6 wk and MLN4924- vs. DMSO-treated groups (B and C; *P < 0.05, **P < 0.01, and ***P < 0.001; n.s., not significant).

Fig. 6.

MLN treatment reduces inflammatory protein expression in vivo. (A) Male Apoe^−/− mice consumed a high-fat diet for 12 wk; in the middle of the diet, mice were injected with MLN4924 or DMSO control solvent for 6 wk. TNF-α and IL-10 levels in serum (n = 8–9). (B–D) Apoe^−/− mice pretreated with MLN4924 or DMSO control solution were injected i.p. with LPS (B). After 24 h, IL-6 levels were measured in serum (C; n = 8–10), and gene expression normalized to 18S was analyzed in macrophages isolated from the peritoneum (D; n = 5–9). (A–D) Data represent means ± SD. t test or Mann–Whitney test was performed: one-tailed (Il-6 in D) or two-tailed (others) as appropriate based on in vitro data (*P < 0.05 and **P < 0.01).