High-mobility group box 1 is a novel deacetylation target of Sirtuin1

Abstract

High-mobility group box 1 (HMGB1) undergoes acetylation, nuclear-to-cytoplasmic translocation, and release from stressed kidneys, unleashing a signaling cascade of events leading to systemic inflammation. Here, we tested whether the deacetylase activity of Sirtuin1 (SIRT1) participates in regulating nuclear retention of HMGB1 to ultimately modulate damage signaling initiated by HMGB1 secretion during stress. When immunoprecipitated acetylated HMGB1 was incubated with SIRT1, HMGB1 acetylation decreased by 57%. Proteomic analysis showed that SIRT1 deacetylates HMGB1 at four lysine residues (55, 88, 90, and 177) within the proinflammatory and nuclear localization signal domains of HMGB1. Genetic ablation or pharmacological inhibition of SIRT1 in endothelial cells increased HMGB1 acetylation and translocation. In vivo, deletion of SIRT1 reduced nuclear HMGB1 while increasing its acetylation and release into circulation during basal and ischemic conditions, causing increased renal damage. Conversely, resveratrol pretreatment led to decreased HMGB1 acetylation, its nuclear retention, decreased systemic release, and reduced tubular damage. Thus, a vicious cycle is set into motion in which the inflammation-induced repression of SIRT1 disables deacetylation of HMGB1, facilitates its nuclear-to-cytoplasmic translocation, and systemic release, thereby maintaining inflammation.

Conflict of interest statement

Disclosure: All the authors declared no competing interests.

Figures

Figure 1. Total and acetylated HMGB1 in HUVEC treated for 4 hours with LPS and SIRT1 inhibitor (SI)

HUVEC treated with SI and LPS for 4 hours demonstrated enhanced translocation (A) and acetylation (B) under basal conditions and when treated with LPS. However, translocated and acetylated HMGB1 did not differ between SI and LPS + SI treatments, indicating lack of additive effect. *p≤0.05 vs. control; n=6 panel A; n=3-5 panel B.

Figure 2. HUVEC stained for HMGB1 after 24 hours of LPS and SIRT1 inhibitor (SI) treatment

(A) HMGB1 immunocytochemistry and (B) quantification of fluorescence intensity demonstrated that HUVEC treated with SIRT1 inhibitor (SI) and LPS for 24 hours showed enhanced translocation from the nucleus to the cytoplasm, as compared to control cells. HMGB1 did not differ between SI and LPS + SI treatment, suggesting the lack of additive effect. *p

Figure 3. HUVEC transfected with HMGB1-GFP plasmid treated for 24 hours with SIRT1 inhibitor (SI) with and without LPS

(A) Fixed transfected cells imaged for HMGB1-GFP translocation and (B) quantification of fluorescence intensity demonstrated that SIRT1 inhibitor increased HMGB1 nuclear-to-cytoplasmic translocation, as indicated by increased GFP fluorescence in the cytoplasm. Administration of LPS also promoted HMGB1 translocation to the cytoplasm. LPS and SIRT1 inhibitor co-administration did not further enhance HMGB1 cytosolic fluorescence, suggesting that both activate similar mechanisms. At longer times, LPS and SIRT1 inhibitor treatment increased HMGB1-GFP fluorescence in the nucleus, suggesting increased protein expression. *p

Figure 4. SIRT1 deacetylation of HMGB1

Immunoprecipitated acetylated (Ac) HMGB1 was combined with recombinant SIRT1 for 60 minutes and levels of acetylated HMGB1 were quantified by immunoblotting. SIRT1 deacetylated HMGB1 by 49% within 1 hour. *p>0.05 vs. untreated; n=7.

Figure 5. Mass spectrometry analysis of the lysine residues in HMGB1 that are deacetylated by SIRT1

Purified acetylated HMGB1 was incubated with SIRT1 and cofactors at 37°C for 1 hour. Subsequently, HMGB1 was examined by mass spectrometry. HMGB1 in samples was proteolytically fragmented with trypsin prior to mass spectrometry analysis, as detailed in Methods. Acetylated lysine residues appeared on the MS/MS spectra as peaks at 126 m/z. Before SIRT1 treatment, lysine residues 55, 88, 90 and 177 were acetylated (A), however, after SIRT1 treatment these residues were no longer acetylated (B), indicating SIRT1 deacetylation.

Figure 6. Illustration of the lysine residues deacetylated by SIRT1, as determined by mass spectrometry analysis

(A) Mass spectrometry analysis revealed lysine residues (K) at positions 55, 88, 90 and 177 (highlighted in red text) within the 215 amino acid sequence of HMGB1 are deacetylated by SIRT1. (B) The positions of the deacetylated residues in the HMG Box A and Box B are shown. Lys55 is in HMG Box A adjacent to Cys45. Lysine residues 88 and 90 are in the pro-inflammatory cytokine domain. Lys177 is in the NLS-2 and RAGE-binding domain. (C) The 3-D protein structure of HMGB1 shows the location of SIRT1 deacetylated lysines (highlighted in blue) in relation to hydrophobic residues (highlighted in yellow), including the presence of large bulky aromatic tryptophan and phenylalanine residues (highlighted in red) adjacent to Lys55 (D) and Lys88 and Lys90 (E), respectively. The 3-D protein structure for the entire HMGB1 for positioning of Lys177 is currently not available. References for the HMGB1 3-D protein structure are in the Methods section.

Figure 7. Nuclear (A) and cytoplasmic (B) HMGB1 in SIRT1 Flox/Flox and SIRT1 endothelial -/- cell primary cultures from the aorta treated with LPS for 4 hours

SIRT1 endothelial -/- cells demonstrated enhanced cytoplasmic translocation under basal conditions and LPS treatment. Nuclear Western blot images for HMGB1 were obtained from different parts of the same gel. *p≤0.05 vs. SIRT1 Flox/Flox; #p<0.05 vs. SIRT1 endothelial -/-; †p<0.10 vs. SIRT1 Flox/Flox; n=4.

Figure 8. Nuclear and plasma HMGB1 in endothelial SIRT1 endothelial -/- mice after renal IRI

Mice were subjected to 30 minutes of bilateral renal IRI and examined for HMGB1 translocation 1 hour after reperfusion. As measured by ELISA and immunobloting, the SIRT1 endothelial -/- cells resulted in reduced nuclear HMGB1 (A) and increased proportion of nuclear acetylated HMGB1 (B), as compared to SIRT1 Flox/Flox mice. Total HMGB1 (C) and acetylated HMGB1 (D) increased in the circulation after renal IRI in SIRT1 Flox/Flox, heterozygous SIRT1 endothelial +/- and SIRT1 endothelial -/- mice with the most robust increase occurring in endothelial SIRT1 endothelial -/- mice. Data in figures lacking Western blot images were obtained by ELISA assay. In Fig 9B, acetylated HMGB1 was quantified by Western blot, while total HMGB1 was quantified by ELISA. *p#p<0.05 vs. SIRT1 Flox/Flox IRI; ˆp<0.01 vs. SIRT1 endothelial -/-; **P<0.05 vs. SIRT1 endothelial -/- IRI; n=4-5.

Figure 9. FVB/NJ mice pre-treated with SIRT1 activator resveratrol prior to renal IRI

Pretreatment with resveratrol prior to bilateral renal IRI resulted in improved HMGB1 nuclear (A) retention 1 hour after IRI. IRI enhanced levels of acetylated HMGB1 in the cytoplasm (B) 1 hour after IRI, but pretreatment with resveratrol normalized HMGB1 acetylation. Total HMGB1 (C) and acetylated HMGB1 (D) increased in the circulation 1 hour after IRI, but were significantly decreased by resveratrol. HMGB1 was elevated in the urine (E) 24 hours after IRI, but was reduced by resveratrol pretreatment. Data in figures lacking Western blot images were obtained by ELISA assay. In Fig 9B, acetylated HMGB1 was quantified by Western blot while total HMGB1 was quantified by ELISA. *p≤0.05 vs. all groups; #p<0.05 vs. IRI; †p<0.10 vs. IRI; n=3-5.

Figure 10. Comparison of the effects of resveratrol pretreatment in SIRT1 Flox/Flox versus SIRT1 endothelial -/- mice

Similar to the affect observed in FVB/NJ, 1 hour after a 30 minute episode of bilateral renal IRI, nuclear levels of HMGB1 were reduced (A) while circulating HMGB1 was enhanced (B) in SIRT1 Flox/Flox mice, an effect that was potentiated when SIRT1 was ablated in endothelial cells. While resveratrol pretreatment prevented the reduction of nuclear HMGB1 during IRI, circulating levels of HMGB1 remained elevated. Bilateral IRI resulted in a rise in BUN in both SIRT1 Flox/Flox and SIRT1 endothelial -/- mice 24 hours after IRI (C). There was a numerical (but statistically insignificant) difference in BUN levels between SIRT1 Flox/Flox and SIRT1 endothelial -/mice after bilateral IRI and resveratrol pretreatment. *p≤0.05 vs. SIRT1 Flox/Flox IRI; ˆp†p<0.05 vs. SIRT1 endothelial -/- control; #p<0.10 vs. SIRT1 Flox/Flox IRI + resveratrol; n=3-5.

Figure 11. Histological analysis of renal damage in SIRT1 Flox/Flox and SIRT1 endothelial -/- mice 24 hours after a 30 minute episode of bilateral renal IRI

SIRT1 endothelial -/- mice exhibited worse renal damage after IRI, including enhanced tubular epithelial cell necrosis, loss of brush border and cast formation, as indicated by hemotoxylin and eosin and periodic acid-shift staining (A) and quantified for pathological score (B), as described by Conger et al. Resveratrol pretreatment attenuated necrosis, but had negligible effects on all other examined parameters. *p#p<0.10 vs. resveratrol (same group); n=3. All images are 400x magnification. Scale bar represents 50 μm.

Figure 12. Histological analysis of renal damage in SIRT1 Flox/Flox and SIRT1 endothelial -/- mice 5 days after a 30 minute episode of unilateral renal IRI

SIRT1 endothelial -/- mice exhibited worse renal damage after IRI in both the ischemic and contralateral control kidney, as indicated by hematoxylin and eosin, periodic acid-shift, and Masson's trichrome staining (A). Quantification analyses of pathological damage (B) indicated the ischemic kidney exhibited enhanced necrosis, brush border loss, cast formation, leuokocyte infiltration and fibrosis, while the contralateral control kidney in SIRT endothelial -/- also demonstrated significant necrosis, brush border loss, tubular dilation and fibrosis. #p<0.05 vs. SIRT1 Flox/Flox control; **p≤0.05 vs. SIRT1 Flox/Flox IRI; †p<0.10 vs. SIRT1 Flox/Flox control; *p<0.10 vs. SIRT1 Flox/Flox IRI; n=3. All images are 400x magnification. Scale bar represents 50 μm.