A two-hit model of sepsis plus hyperoxia causes lung permeability and inflammation

Abstract

Mouse models of acute lung injury (ALI) have been instrumental for studies of the biological underpinnings of lung inflammation and permeability, but murine models of sepsis generate minimal lung injury. Our goal was to create a murine sepsis model of ALI that reflects the inflammation, lung edema, histological abnormalities, and physiological dysfunction that characterize ALI. Using a cecal slurry (CS) model of polymicrobial abdominal sepsis and exposure to hyperoxia (95%), we systematically varied the timing and dose of the CS injection, fluids and antibiotics, and dose of hyperoxia. We found that CS alone had a high mortality rate that was improved with the addition of antibiotics and fluids. Despite this, we did not see evidence of ALI as measured by bronchoalveolar lavage (BAL) cell count, total protein, C-X-C motif chemokine ligand 1 (CXCL-1) or by lung wet:dry weight ratio. Addition of hyperoxia [95% fraction of inspired oxygen ([Formula: see text])] to CS immediately after CS injection increased BAL cell counts, CXCL-1, and lung wet:dry weight ratio but was associated with 40% mortality. Splitting the hyperoxia treatment into two 12-h exposures (0-12 h and 24-36 h) after CS injection increased survival to 75% and caused significant lung injury compared with CS alone as measured by increased BAL total cell count (92,500 vs. 240,000, P = 0.0004), BAL protein (71 vs. 103 µg/mL, P = 0.0030), and lung wet:dry weight ratio (4.5 vs. 5.5, P = 0.0005), and compared with sham as measured by increased BAL CXCL-1 (20 vs. 2,372 pg/mL, P < 0.0001) and histological lung injury score (1.9 vs. 4.2, P = 0.0077). In addition, our final model showed evidence of lung epithelial [increased BAL and plasma receptor for advanced glycation end products (RAGE)] and endothelial (increased Syndecan-1 and sulfated glycosaminoglycans) injury. In conclusion, we have developed a clinically relevant mouse model of sepsis-induced ALI using intraperitoneal injection of CS, antibiotics and fluids, and hyperoxia. This clinically relevant model can be used for future studies of sepsis-induced ALI.

Keywords: ARDS; acute lung injury; acute respiratory distress syndrome; hyperoxia; sepsis.

Figure 1.

Effect of antibiotic and fluid timing on severity of illness and lung injury. Mice were treated with 2.0 mg/g of CS or sham and given antibiotics and fluids starting at either 8 or 12 h after CS injection (n = 10–18/group). Mice were then harvested at 48 h after CS injection. A: weight change did not differ between septic groups. Sepsis score (B) and survival rate (C) were lowest (most severe) in mice that received CS without Abx/Fl but did not differ between the Abx/Fl-receiving groups. BAL total cell count (D) and BAL protein (F) were not significantly higher than sham in any group. BAL CXCL-1 (E) and lung wet:dry ratio (G) were highest in mice that received CS without Abx/Fl. [Statistical analysis: mixed-effects model + Tukey’s post hoc (each timepoint), median + 95% CI (A–B); log-rank (Mantel-Cox), survival proportions (C); Kruskal–Wallis + Dunn’s mc, median + 95% CI (D–G)]. Each point represents an individual animal. BAL, bronchoalveolar lavage; CI, confidence interval; CS, cecal slurry; mc, multiple comparisons. Pink * = Sham; RA vs. CS; RA. Green # = Sham; RA vs. CS; [0-24]. D. Purple $ = Sham; RA vs. CS; [12-36].

Figure 2.

Effect of hyperoxia duration and timing on severity of illness and lung injury. Mice were treated with 2.4 mg/g of CS and given antibiotics and fluids starting at 12 h after CS injection. Mice were exposed to room air or 95% O₂ for 6 h starting at 6 and/or 24 h after CS injection (n = 2–12/group). Mice were then harvested at 30 h after CS injection. Weight change (A) and sepsis score (B) did not differ between septic groups. Survival rate (C) was lowest in mice (n = 3) that received 2.4 mg/g CS plus HO (6–12 h). BAL total cell count (D), BAL protein (F), and lung wet:dry ratio (G) did not significantly differ from sham control for all septic groups. BAL CXCL-1 (E) was significantly elevated in the CS plus HO (6–12 h, 24–30 h) group (n = 12). [Statistical analysis: mixed-effects model + Tukey’s mc (each timepoint), median + 95% CI (A–B); log-rank (Mantel-Cox), survival proportions (C); Kruskal–Wallis + Dunn’s mc, median + 95% CI (D–G)]. BAL, bronchoalveolar lavage; CI, confidence interval; CS, cecal slurry; HO, hyperoxia; mc, multiple comparisons; RA, room air. * = Sham; RA vs. CS; RA. D. Purple $ = Sham; RA vs. CS; [12-36]. L. Purple % = Sham; RA vs. CS; [0-12, 24-36].

Figure 3.

Effect of hyperoxia timing on severity of illness and lung injury. Mice were treated with 2.4 mg/g of CS (n = 6–46) and given antibiotics and fluids starting at 12 h after CS injection. Mice were exposed to room air or 95% O₂ for 24 h consecutively starting at 0 or 12 h after CS injection, or 24 h cumulatively in 12-h doses starting at 0 and 24 h after CS injection. Weight change (A) and sepsis score (B) did not differ between septic groups. C: survival rate was lowest in CS plus HO (0–24h) and CS plus HO (0–12h, 24–36h) groups (n = 35–46/group). BAL total cell count (D) and BAL CXCL-1 (E) were significantly elevated in all CS plus HO groups (n = 14–22/group). F: BAL protein was significantly elevated in the CS plus HO (12–26 h) and CS plus HO (0–12 h, 24–36 h) groups (n = 15–27/group). G: lung wet:dry ratio was significantly elevated in the CS plus HO (0–24 h) and CS plus HO (0–12 h, 24–36 h) groups (n = 7–17/group). [Statistical analysis: mixed-effects model + Tukey’s mc (each timepoint), median + 95% CI (A–B); log-rank (Mantel-Cox), survival proportions (C); Kruskal–Wallis + Dunn’s mc, median + 95% CI (D–G)]. BAL, bronchoalveolar lavage; CI, confidence interval; CS, cecal slurry; HO, hyperoxia; mc, multiple comparisons; RA, room air. Pink * = Sham; RA vs. CS; RA. Green # = Sham; RA vs. CS; [0-24]. D. Purple $ = Sham; RA vs. CS; [12-36]. L. Purple % = Sham; RA vs. CS; [0-12, 24-36]. Green + = CS; RA vs. CS; [0-24]. D. Purple & = CS; RA vs. CS; [12-36]. L. Purple ∧ = CS; RA vs. CS; [0-12, 24-36].

Figure 4.

Experimental design and lung cytokine levels for split hyperoxia and CS two-hit model. Mice were treated with 2.4 mg/g of CS and given antibiotics and fluids starting at 12 h after CS or vehicle injection. Mice were exposed to room air or 95% O₂ for 24 h cumulatively in 12-h doses starting at 0 and 24 h after CS or vehicle injection. Mice were then harvested after 36 h (n = 3–18/group). A: experimental design of split hyperoxia and CS two-hit model. Lung CXCL-1 (B), IL-6 (C), IL-10 (D), and TNF-α (E) did not significantly differ between septic groups. [Statistical analysis: Kruskal–Wallis + Dunn’s mc, median + 95% CI (B–E)]. CI, confidence interval; CS, cecal slurry; HO, hyperoxia; ns, not significant. Black = sham; gray = HO; pink = CS; purple = CS + HO. Pink * = Sham; RA vs. CS; RA. L. Purple % = Sham; RA vs. CS; [0-12, 24-36].

Figure 5.

Effect of CS and hyperoxia on lung injury histology score. Mice were treated with 2.4 mg/g of CS and given antibiotics and fluids starting at 12 h after CS injection. Mice were exposed to room air or 95% O₂ for 24 h cumulatively in 12-h doses starting at 0 and 24 h after CS injection (n = 3–12/group). Sham (A), hyperoxia alone (B), CS alone (C), and CS plus hyperoxia (0–12 h, 24–36 h) (D) lungs were stained with H&E and imaged at ×20 magnification. E: lung histology score was significantly elevated for CS plus HO (0–12 h, 24–36 h). [Statistical analysis: Kruskal–Wallis + Dunn’s mc, median + 95% CI (E)]. CI, confidence interval; CS, cecal slurry; H&E, hematoxylin-eosin; HO, hyperoxia. Black = sham, grey = HO, pink = CS, purple=CS+HO. L. Purple % = Sham; RA vs. CS; [0-12, 24-36].

Figure 6.

Effect of CS and hyperoxia on BAL and plasma injury markers. Mice were treated with 2.4 mg/g of CS and given antibiotics and fluids starting at 12 h after CS injection. Mice were exposed to room air or 95% O₂ for 24 h cumulatively in 12-h doses starting at 0 and 24 h after CS injection (n = 3–17). BAL (A) and plasma (B) RAGE were significantly elevated in the CS plus HO group. Syndecan-1 (C) and sulfated GAGs (D) did not differ between septic groups. [Statistical analysis: Kruskal–Wallis + Dunn’s mc, median + 95% CI (A–F)]. BAL, bronchoalveolar lavage; CI, confidence interval; CS, cecal slurry; GAGs, glycosaminoglycans; HO, hyperoxia; RAGE, receptor for advanced glycation end products. Black = sham, grey = HO, pink = CS, purple = CS+HO. * = Sham; RA vs. CS; RA. L. Purple % = Sham; RA vs. CS; [0-12, 24-36].