Redox modification of cysteine residues regulates the cytokine activity of high mobility group box-1 (HMGB1)

Abstract

High mobility group box 1 (HMGB1) is a nuclear protein with extracellular inflammatory cytokine activity. It is released passively during cell injury and necrosis, and secreted actively by immune cells. HMGB1 contains three conserved redox-sensitive cysteine residues: C23 and C45 can form an intramolecular disulfide bond, whereas C106 is unpaired and is essential for the interaction with Toll-Like Receptor (TLR) 4. However, a comprehensive characterization of the dynamic redox states of each cysteine residue and of their impacts on innate immune responses is lacking. Using tandem mass spectrometric analysis, we now have established that the C106 thiol and the C23-C45 disulfide bond are required for HMGB1 to induce nuclear NF-κB translocation and tumor necrosis factor (TNF) production in macrophages. Both irreversible oxidation to sulphonates and complete reduction to thiols of these cysteines inhibited TNF production markedly. In a proof of concept murine model of hepatic necrosis induced by acetaminophen, during inflammation, the predominant form of serum HMGB1 is the active one, containing a C106 thiol group and a disulfide bond between C23 and C45, whereas the inactive form of HMGB1, containing terminally oxidized cysteines, accumulates during inflammation resolution and hepatic regeneration. These results reveal critical posttranslational redox mechanisms that control the proinflammatory activity of HMGB1 and its inactivation during pathogenesis.

Figure 1

Characterization of redox-dependent modifications of cysteine residues in HMGB1 using mass spectrometry. Molecule A: mass spectrometric characterization of molecule A on C106 (A-1), C23 (A-2) and C45 (A-3). A-1: Peptide-containing HMGB1 amino acids 97–112 with a thiol capped NEM adduct indicating reduced C106. A-2: Peptide-containing HMGB1 amino acids 13–24 following DTT reduction of a disulfide bond and subsequent adduction of C23 with d₅NEM. A-3: Peptide-containing HMGB1 amino acids 45–48 following DTT reduction of a disulfide bond and subsequent adduction of C45 with d₅NEM. Molecule B: mass spectrometric characterization of molecule B on C106 (B-1), C23 (B-2) and C45 (B-3). B-1: Peptide-containing HMGB1 amino acids 97–112 containing a C106 as a sulphonic acid. B-2: Peptide-containing HMGB1 amino acids 13–24 containing a C23 as a sulphonic acid. B-3: Peptide-containing HMGB1 amino acids 45–48 containing a C45 as a sulphonic acid. Molecule C: mass spectrometric characterization of molecule C on C23 (C-1) and C45 (C-2). C-1: Mass spectrometric characterization of C23. Peptide-containing HMGB1 amino acids 13–24 with a thiol-capped NEM adduct indicating reduced C23. C-2: Peptide-containing HMGB1 amino acids 45–48 with a thiol-capped NEM adduct indicating reduced C45. Peptide sequences and b and y ions are indicated on each trace as required. MS/MS traces are representative of three independent investigations.

Figure 1

Characterization of redox-dependent modifications of cysteine residues in HMGB1 using mass spectrometry. Molecule A: mass spectrometric characterization of molecule A on C106 (A-1), C23 (A-2) and C45 (A-3). A-1: Peptide-containing HMGB1 amino acids 97–112 with a thiol capped NEM adduct indicating reduced C106. A-2: Peptide-containing HMGB1 amino acids 13–24 following DTT reduction of a disulfide bond and subsequent adduction of C23 with d₅NEM. A-3: Peptide-containing HMGB1 amino acids 45–48 following DTT reduction of a disulfide bond and subsequent adduction of C45 with d₅NEM. Molecule B: mass spectrometric characterization of molecule B on C106 (B-1), C23 (B-2) and C45 (B-3). B-1: Peptide-containing HMGB1 amino acids 97–112 containing a C106 as a sulphonic acid. B-2: Peptide-containing HMGB1 amino acids 13–24 containing a C23 as a sulphonic acid. B-3: Peptide-containing HMGB1 amino acids 45–48 containing a C45 as a sulphonic acid. Molecule C: mass spectrometric characterization of molecule C on C23 (C-1) and C45 (C-2). C-1: Mass spectrometric characterization of C23. Peptide-containing HMGB1 amino acids 13–24 with a thiol-capped NEM adduct indicating reduced C23. C-2: Peptide-containing HMGB1 amino acids 45–48 with a thiol-capped NEM adduct indicating reduced C45. Peptide sequences and b and y ions are indicated on each trace as required. MS/MS traces are representative of three independent investigations.

Figure 2

Effect of oxidation agents on HMGB1-induced TNF release. The time dependent inhibitory effect of H₂O₂ exposure on the TNF-stimulating activity of HMGB1 in (A) RAW 267.7 cells and (B) primary human macrophages. (C) The effect of H₂O₂ exposure on the TNF-stimulating activity of LPS in RAW 267.7 cells. Cells were cultured in 96-well plates and were stimulated with HMGB1 or LPS with or without exposure to 50 mmol/L H₂O₂ for 0–120 min. After16 h, TNF released into the cell culture supernatant was measured by ELISA. N = 6–8. *P < 0.05 versus HMGB1 alone without H₂O₂ exposure. (D) The effect of mercury (Hg) on TNF release from RAW 264.7 cells induced by recombinant HMGB1 prepared in the absence of DTT. RAW 264.7 cells were cultured in 96-well plates and exposed to HMGB1 or Hg-HMGB1 over a range of concentrations (0–5 μg/mL). After 16 h, TNF released was measured. N = 3. *P < 0.05 versus HMGB1.

Figure 3

Redox dependent effect on HMGB1-induced NF-κB activation in macrophages. Western analysis of the nuclear localization of the NF-κB p65 subunit following the stimulation of cultured macrophages for 1 h with 5 μg/mL of HMGB1 or 4 ng/mL of LPS. HMGB1 was exposed to either H₂O₂ (50 mmol/L, 120 min) or DTT (5 mmol/L, 120 min) prior to the assay. For HMGB1 preparations that had included DTT, HMGB1 was then oxidized with 50 μmol/L H₂O_2. β actin expression was also measured and used as a loading control. All HMGB1 preparations were characterized by LC-MS/MS prior to cell treatment with respect to cysteine redox status as described in Figure 1. Data shown are representative of three independent experiments. A, control (PBS); B, HMGB1 (molecule B + H₂O₂); C, HMGB1 (molecule C + DTT); D, LPS; E, HMGB1 (molecule A); F, HMGB1 (molecule C + H₂O₂).

Figure 4

Effect of reducing agents on HMGB1-induced cytokine release. The (A) dose dependent and (B) time dependent effect of DTT on the TNF-stimulating activity of HMGB1 in RAW 264.7 cells. (C) The effect of DTT on the TNF-stimulating activity of LPS in RAW 264.7 cells. HMGB1 or LPS was incubated with the reducing agent DTT at (A) various concentrations as shown or with 5 mmol/L DTT at the times indicated at room temperature. Fresh DTT pretreated HMGB1 or LPS was then added to RAW 264.7 cells. After 16 h, TNF released into the cell culture supernatant was measured by ELISA. N = 3. *P < 0.05 versus HMGB1 alone without DTT pretreatment.

Figure 5

The requirement for a disulfide bond between C23-C45 for the cytokine-stimulating activity of HMGB1. (A) Effects of mild H₂O₂ exposure (50 μmol/L, 120 min) on the TNF stimulating activity of HMGB1 prepared with DTT. (B) Effects of an A45 mutation of HMGB1 on TNF release from RAW 264.7 cells induced by A45 HMGB1 compared with HMGB1 prepared in the absence of DTT. RAW 264.7 cells were incubated with HMGB1 over a range of concentrations (0 to 10 μg/mL). After 16 h, TNF released into the cell culture supernatant was measured by ELISA. N = 3. *P < 0.05 versus HMGB1. All HMGB1 preparations were characterized by LC-MS/MS prior to cell treatment with respect to cysteine redox status.

Figure 6

Mass spectrometric characterization of circulating HMGB1 and its association with hepatic inflammatory cell recruitment during acetaminophen hepatotoxicity. Histo-logical characterization of hepatic changes induced by acetaminophen (530 mg/kg) over 5, 15 and 24 h in mice. Necrotic hepatocytes are indicated with an arrow and infiltrating inflammatory cells are highlighted with an arrow head. Data shown are representative of six animals per treatment group.