Therapeutic targeting of HMGB1 during experimental sepsis modulates the inflammatory cytokine profile to one associated with improved clinical outcomes

Abstract

Sepsis remains a significant health burden and a major clinical need exists for therapeutics to dampen the excessive and uncontrolled immune activation. Nuclear protein high mobility group box protein 1 (HMGB1) is released following cell death and is a late mediator in sepsis pathogenesis. While approaches targeting HMGB1 have demonstrated reduced mortality in pre-clinical models of sepsis, the impact of HMGB1 blockade on the complex septic inflammatory milieu and the development of subsequent immunosuppression remain enigmatic. Analysis of plasma samples obtained from septic shock patients established an association between increased HMGB1 and non-survival, higher APACHE II scores, and increased pro-inflammatory cytokine responses. Pre-clinically, administration of neutralising ovine anti-HMGB1 polyclonal antibodies improved survival in murine endotoxaemia and caecal ligation and puncture-induced sepsis models, and altered early cytokine profiles to one which corresponded to patterns observed in the surviving patient cohort. Additionally, anti-HMGB1 treated murine sepsis survivors were significantly more resistant to secondary bacterial infection and exhibited altered innate immune cell phenotypes and cytokine responses. These findings demonstrate that anti-HMGB1 antibodies alter inflammation in murine sepsis models and reduce sepsis mortality without potentiating immunosuppression.

Figure 1

Plasma HMGB1 concentrations and disease severity and mortality in patients with septic shock. Plasma samples from ICU patients with diagnoses consistent with septic shock were taken twice daily (morning and afternoon) for the length of ICU stay. (A) APACHE II scores in survivors and non-survivors were compared via two-tailed t test. (B) HMGB1 was quantified in patient samples by ELISA and regression analyses performed to compare plasma HMGB1 at 24 hours post ICU admission and APACHE II score. (C) Mean kinetic plasma HMGB1 levels in survivors versus non-survivors were compared by paired two-way t-test. Significance is denoted as thus: *P < 0.05, **P < 0.01.

Figure 2

Mortality and elevated HMGB1 are associated with a proinflammatory serum cytokine profile in a cohort of septic shock patients. Plasma samples from septic shock patients were stored and analysed via multiplex analysis. A panel of cytokines (IL-6, MCP-1, IL-10, IFNγ, TNFα, IL-1β, IL-4, IL-5, IL-8 and IL-1RA) was assessed and concentrations calculated from samples run in duplicate. (A) Plasma concentrations of individual cytokines between survivors (black) and non-survivors (grey) were compared via paired t-test. Cytokine measurements normalised to the median concentration were used to determine composite cytokine scores (CCS). (B) Total CCS indicating the total magnitude of cytokine levels was calculated by summation of all ten normalised cytokine scores and compared between survivors and non-survivors. The ratio of pro-inflammatory (IL-6, MCP-1, IFNγ, TNFα, IL-1β, IL-5 and IL-8) to anti-inflammatory (IL-10, IL-4 and IL-1RA) signalling was compared between samples with a low and high HMGB1 content (C) and between survivors and non-survivors (D). Plasma cytokine concentrations and ratio between survivors and non-survivors were compared via paired t-test and low vs high HMGB1 content samples were compared via unpaired t-test. Data are depicted as mean ± SEM, with significance is denoted as thus: *P < 0.05, **P < 0.01.

Figure 3

Ovine anti-HMGB1 pAb therapy improves survival in an in vivo model of murine endotoxaemia. Adult male and female B6 mice were prophylactically administered 20 mg/kg anti-HMGB1 pAb (n = 7), control pAb (n = 7) or PBS (n = 7) via IP injection one hour before IP injection of 20 mg/kg LPS (A&C). In a therapeutic model, mice were administered 20 mg/kg anti-HMGB1 pAb (n = 6), control pAb (n = 4) or PBS (n = 7) one hour after LPS administration (B&D). Experimental mice were monitored over twenty-four hours for clinical score and euthanised if predetermined humane endpoints were reached. Survival (A,B) and mean clinical score (C,D) of groups is represented. Survival curves were compared to controls using the Mantel-Cox test and clinical score data compared to PBS control via two-way ANOVA, with significance denoted as: *P < 0.05, **P < 0.01.

Figure 4

Therapeutic administration of ovine anti-HMGB1 pAb improves survival from CLP sepsis. Mice were subjected to sham-surgery alone (n = 4) or CLP surgery (n = 4) and serum samples from sham or CLP-surgery mice were assayed for HMGB1 via ELISA (A). CLP mice (n = 9–11) were administered PBS, control antibody or anti-HMGB1 pAb (25 mg/kg) via IP injection eight hours following surgery. Mice were monitored and euthanised if predetermined humane endpoints were reached. Mean ± SEM of clinical score (B) and weight (C) and survival curves of the groups (D) are depicted. Sequential measurements were compared via two-way ANOVA and survival of anti-HMGB1 treated mice was compared to control groups using the Mantel Cox test with significance denoted as: *P > 0.05, a*P > 0.05 compared against PBS control, b*P > 0.05 compared against control Ab.

Figure 5

Ovine anti-HMGB1 pAb-treated mice display cytokine profiles associated with improved clinical outcomes in the early phase of CLP-induced sepsis. Blood samples from mice (n = 3–5) subjected to sham-surgery (pale grey area) or CLP-induced sepsis and treated with PBS (unfilled symbols), control pAb (25 mg/kg; black symbols) or anti-HMGB1 pAb (25 mg/kg; grey symbols) were analysed for cytokine content using a multiplex platform. Mean + SEM of individual cytokine measurements are depicted. (A). Proinflammatory (IL-2, IL-5, IL-6, IL-12, IFNγ, GM-CSF and TNFα) to anti-inflammatory (IL-4, IL-10 & IL-13) CCS were calculated for each sample, mean + SEM group CCS normalised to median values are depicted in (B). Levels of cytokines were compared using two-tailed T-tests, with significance denoted as thus: *P < 0.05.

Figure 6

Prior anti-HMGB1 pAb treatment alters the inflammatory response and survival to secondary bacterial challenge in mice that have survived CLP-induced sepsis. Mice (n = 6–12) were subjected to sham-surgery or mild CLP and allowed to recover before IN instillation of P. aeruginosa (2.5 × 10⁷ CFU) or PBS on Day 4 (indicated by dashed line). Mice were treated 8 hours following surgery with anti-HMGB1 or control pAbs (20 mg/kg) or PBS and administered maintenance pAb doses (10 mg/kg) or PBS on days 3 and 6. Mice were monitored for disease symptoms and euthanised at predetermined humane endpoints. Mean - SEM weight loss (A; humane weight loss endpoint indicated by horizontal dashed line) and survival (B) and of mice are depicted. Serum samples collected from mice (n = 4–5) immediately prior to bacterial challenge and 24 hours after challenge were analysed for IL-6, IL-10, IL-12 or TNFα content via ELISA (C). Survival curves were compared via Mantel-Cox analysis, and changes in cytokines before and after treatment were compared via paired t-test. Cytokine changes between groups were compared via two-way matched pairs ANOVA, with significance denoted as thus: *P < 0.05, **P < 0.01.

Figure 7

Anti-HMGB1 pAb treatment alters macrophage and DC cytokine expression post survival from CLP-induced sepsis. Peritoneal cells were harvested four days following CLP surgery and treatment with either PBS (n = 4), control ovine pAb (n = 4) or anti-HMGB1 pAb (n = 5) and evaluated via flow cytometry. Frequency of large peritoneal macrophages (LPM; CD11b^hi, F4/80^hi, Gr-1⁺) and small peritoneal macrophages (SPM; CD11b^lo, F4/80^lo, Gr-1⁻) compared to normal mice (n = 2) are expressed (A; Mean + SEM). Adherent macrophages from peritoneal lavages were stimulated with LPS (100 ng/mL) for 18 hours and IL-6, IL-10, IL-12 and TNFα content in the supernatant was evaluated via ELISA. Mean + SEM expression for each treatment relative to unstimulated wells is expressed (B). Bone marrow cells harvested from Day 4 septic or sham-surgery mice were cultured in the presence of GM-CSF to generate bone-marrow derived dendritic cells (BMDCs). Harvested BMDC’s were plated and cultured in the presence of LPS (100 ng/mL); Mean + SEM IL-10 and IL-12 cytokine content of resulting supernatants is expressed (C). Mean values of individual groups were compared via two-tailed t tests, with significance denoted as thus: **P < 0.01, *P < 0.05.