Liraglutide pretreatment attenuates sepsis-induced acute lung injury

Abstract

There are no effective targeted therapies to treat acute respiratory distress syndrome (ARDS). Recently, the commonly used diabetes and obesity medications, glucagon-like peptide-1 (GLP-1) receptor agonists, have been found to have anti-inflammatory properties. We, therefore, hypothesized that liraglutide pretreatment would attenuate murine sepsis-induced acute lung injury (ALI). We used a two-hit model of ALI (sepsis+hyperoxia). Sepsis was induced by intraperitoneal injection of cecal slurry (CS; 2.4 mg/g) or 5% dextrose (control) followed by hyperoxia [HO; fraction of inspired oxygen ([Formula: see text]) = 0.95] or room air (control; [Formula: see text] = 0.21). Mice were pretreated twice daily with subcutaneous injections of liraglutide (0.1 mg/kg) or saline for 3 days before initiation of CS+HO. At 24-h post CS+HO, physiological dysfunction was measured by weight loss, severity of illness score, and survival. Animals were euthanized, and bronchoalveolar lavage (BAL) fluid, lung, and spleen tissues were collected. Bacterial burden was assessed in the lung and spleen. Lung inflammation was assessed by BAL inflammatory cell numbers, cytokine concentrations, lung tissue myeloperoxidase activity, and cytokine expression. Disruption of the alveolar-capillary barrier was measured by lung wet-to-dry weight ratios, BAL protein, and epithelial injury markers (receptor for advanced glycation end products and sulfated glycosaminoglycans). Histological evidence of lung injury was quantified using a five-point score with four parameters: inflammation, edema, septal thickening, and red blood cells (RBCs) in the alveolar space. Compared with saline treatment, liraglutide improved sepsis-induced physiological dysfunction and reduced lung inflammation, alveolar-capillary barrier disruption, and lung injury. GLP-1 receptor activation may hold promise as a novel treatment strategy for sepsis-induced ARDS. Additional studies are needed to better elucidate its mechanism of action.NEW & NOTEWORTHY In this study, pretreatment with liraglutide, a commonly used diabetes medication and glucagon-like peptide-1 (GLP-1) receptor agonist, attenuated sepsis-induced acute lung injury in a two-hit mouse model (sepsis + hyperoxia). Septic mice who received the drug were less sick, lived longer, and displayed reduced lung inflammation, edema, and injury. These therapeutic effects were not dependent on weight loss. GLP-1 receptor activation may hold promise as a new treatment strategy for sepsis-induced acute respiratory distress syndrome.

Keywords: GLP-1 receptor activation; acute lung injury; liraglutide; lung edema; lung inflammation.

Graphical abstract

Figure 1.

Effect of liraglutide pretreatment on severity of illness and mortality associated with a 24-h septic insult [cecal slurry (CS; 2.4 mg/g) + hyperoxia (HO; 95% O₂)]. Sepsis score (A; *P < 0.05) and survival rate (B; P < 0.0001) were significantly higher (less severe) in septic mice pretreated with liraglutide compared with saline. n = 46–48. [Statistical analysis: two-way ANOVA (mixed model) + Sidak’s multiple comparisons test (A); log-rank (Mantel-Cox; B)]. Each point represents an individual animal. Data in A were summarized as median ± 95% confidence interval. Data points in B were displayed as survival portions.

Figure 2.

Effect of liraglutide pretreatment on lung cytokine transcript expression at 6-h post septic insult [cecal slurry (CS; 2.4 mg/g) + hyperoxia (HO; 95% O₂)]. Lung tissue IL-1β mRNA (C; *P = 0.0173) expression was significantly lower in septic mice pretreated with liraglutide compared with saline. Liraglutide pretreatment also trended toward lower lung tissue IL-6 (B) and CXCL-1 (D) mRNA expression, but higher IL-10 (E) in septic mice compared with saline. Pretreatment did not affect lung tissue TNF-α (A) mRNA expression in septic mice at this time point. n = 14–19. [Statistical analysis: Mann–Whitney U test (A–E)]. Each point represents an individual animal. Data in A–E were summarized as boxplots, where the box encompassed the interquartile range, error bars encompassed minimum to maximum values, and horizontal line showed median. *P < 0.05. CXCL-1, C-X-C motif chemokine ligand 1; IL-1β, interleukin-1β; IL-6, interleukin-6; IL-10, interleukin-10; TNF-α, tumor necrosis factor-α.

Figure 3.

Effect of liraglutide pretreatment on lung inflammation induced by a septic insult [cecal slurry (CS; 2.4 mg/g) + hyperoxia (HO; 95% O₂)]. BAL total cell counts (A; *P = 0.0021), as well as the concentration of TNF-α (B; *P = 0.0116), IL-6 (C; *P = 0.0026), IL-1β (D; *P = 0.0051), and CXCL-1 (E; *P = 0.0057) were significantly reduced in septic mice pretreated with liraglutide compared with saline. The concentration of BAL IFN-γ (F) was also trending toward being lower in liraglutide-pretreated septic mice compared with saline. Similarly, liraglutide pretreatment in septic mice also trended toward lower MPO activity (H) in lung tissue and BAL neutrophil counts (I) compared with saline. The BAL IL-10 (G) concentration was unaffected by pretreatment. n = 7–14. [Statistical analysis: Mann–Whitney U test (A, H, I); two-tailed unpaired t test on natural log-transformed data (B–H)]. Each point represents an individual animal. Data in A–I were summarized as boxplots, where the box encompassed the interquartile range, error bars encompassed minimum to maximum values, and horizontal line showed median. *P < 0.05. BAL, bronchoalveolar lavage; CXCL-1, C-X-C motif chemokine ligand 1; IFN-γ, interferon gamma; IL-1β, interleukin-1β; IL-6, interleukin-6; IL-10, interleukin-10; MPO, myeloperoxidase; TNF-α, tumor necrosis factor-α.

Figure 4.

Effect of liraglutide on alveolar–capillary injury induced by a septic insult [cecal slurry (CS; 2.4 mg/g) + hyperoxia (HO; 95% O₂)]. Wet-to-dry lung weight ratios (A; *P = 0.0238) as well as BAL protein (B; *P < 0.0001), sulfated GAGs (C; *P = 0.0159), and RAGE (D; *P = 0.0085) were significantly lower in septic mice pretreated with liraglutide compared with saline. n = 7–14. [Statistical analysis: Mann–Whitney U test (A–D)]. Each point represents an individual animal. Data in A–D were summarized as boxplots, where the box encompassed the interquartile range, error bars encompassed minimum to maximum values, and horizontal line showed median. *P < 0.05. BAL, bronchoalveolar lavage; GAGs, glycosaminoglycans; RAGE, receptor for advanced glycation end products.

Figure 5.

Effect of liraglutide pretreatment on lung injury induced by a septic insult [cecal slurry (CS; 2.4 mg/g) + hyperoxia (HO; 95% O₂)]. Total lung injury scores (E; *P < 0.0023) were significantly lower in septic mice pretreated with liraglutide. Representative whole lung sections in septic mice pretreated with saline (A and C) or liraglutide (B and D) used for histology scoring. Lungs were stained with H&E and 20× images of the entire section scanned by the Vanderbilt University Medical Center Digital Histology Shared Resource using the Leica SCN400 Slide Scanner. Individual parameters of lung histology injury score are presented as a table (F). n = 11–17. [Statistical analysis: Mann–Whitney U test (E and F)]. Each point represents an individual animal. Data in E were summarized as boxplots, where the box encompassed the interquartile range, error bars encompassed minimum to maximum values, and horizontal line showed median. *P < 0.05. H&E, hematoxylin-eosin; IQR, interquartile range; RBC, red blood cell.

Figure 6.

Effect of liraglutide pretreatment on liver and kidney injury induced by a septic insult [cecal slurry (CS; 2.4 mg/g) + hyperoxia (HO; 95% O₂)]. Plasma concentrations of ALT (A; *P < 0.0001), AST (B; *P = 0.0005), and BUN (C; *P = 0.0146) were significantly lower in septic mice pretreated with liraglutide compared with saline. Plasma creatinine (D) concentration was unaffected by pretreatment. n = 10–20. [Statistical analysis: Mann–Whitney U test, median + 95% confidence interval (A–D)]. Each point represents an individual animal. Data in A–D were summarized as boxplots, where the box encompassed the interquartile range, error bars encompassed minimum to maximum values, and horizontal line showed median. *P < 0.05. ALT, alanine aminotransferase; AST, aspartate aminotransferase; BUN, blood urea nitrogen.

Figure 7.

Effect of liraglutide pretreatment on systemic inflammation induced by a septic insult [cecal slurry (CS; 2.4 mg/g) + hyperoxia (HO; 95% O₂)]. Plasma concentrations of TNF-α (A), IL-1β (C), and IFN-γ (E) trended toward being lower in liraglutide-pretreated septic mice compared with saline. Pretreatment did not affect plasma concentrations of IL-6 (B), CXCL-1 (D), or IL-10 (F) in septic mice. n = 6–11. [Statistical analysis: two-tailed unpaired t test on natural log-transformed data (A–F)]. Each point represents an individual animal. Data in A–F were summarized as boxplots where the box encompassed the interquartile range, error bars encompassed minimum to maximum values, and horizontal line showed median. CXCL-1, C-X-C motif chemokine ligand 1; IFN-γ, interferon gamma; IL-1β, interleukin-1β; IL-6, interleukin-6; IL-10, interleukin-10; TNF-α, tumor necrosis factor-α.

Figure 8.

The effect of cecal slurry dose on the therapeutic benefits of liraglutide pretreatment for attenuating illness severity, mortality, lung inflammation, alveolar–capillary barrier dysfunction, spleen bacterial burden, and nonpulmonary organ injury induced by a septic insult [cecal slurry (CS; 2.4 mg/g) + hyperoxia (HO; 95% O₂)]. Cecal slurry dose was calculated based on an animal’s initial (before pretreatment) body weight. Sepsis score (A; *P < 0.0001) and survival rate (B; P = 0.0013) were significantly higher (less severe) in septic mice pretreated with liraglutide compared with saline. n = 9–31. Septic mice pretreated with liraglutide also had significantly lower BAL total cell counts (C; *P = 0.0062), BAL IL-1β (D; *P = 0.0041), BAL TNF-α (E; P = 0.0450), and wet-to-dry lung weight ratios (J; *P = 0.0006). Septic mice pretreated with liraglutide also trended toward lower BAL CXCL-1 (F), IFN-γ (G), and RAGE content (L). BAL IL-6 (H), IL-10 (I), and protein (K), as well as percent change in body weight (M) and bacterial burden in spleen (N) were unaffected by pretreatment. [Statistical analysis: two-way ANOVA (mixed model) + Sidak’s multiple comparisons test (A), log-rank (Mantel-Cox; B); Mann–Whitney U test (C, J–N); two-tailed unpaired t test on natural log-transformed data (D–I)]. Each point represents an individual animal. Data in A were summarized as the median ± 95% confidence interval. Data in B were summarized as survival portions. Data in C–M were summarized as boxplots, where the box encompassed the interquartile range, error bars encompassed minimum to maximum values, and horizontal line showed median. BAL, bronchoalveolar lavage; CXCL-1, C-X-C motif chemokine ligand 1; IFN-γ, interferon gamma; IL-1β, interleukin-1β; IL-6, interleukin-6; IL-10, interleukin-10; RAGE, receptor for advanced glycation end products; TNF-α, tumor necrosis factor-α.

Figure 9.

The effect of weight loss induced by caloric restriction or liraglutide for attenuating illness severity, mortality, lung inflammation, alveolar–capillary barrier dysfunction, and bacterial spleen counts induced by a septic insult [cecal slurry (CS; 2.4 mg/g) + hyperoxia (HO; 95% O₂)]. Saline pretreatment was paired with a model of caloric restriction, where animals only had access to food 6–8 h per day. Liraglutide-pretreated animals had free access to food. n = 12–24. Septic mice pretreated with liraglutide also had significantly lower BAL IL-1β (D; *P = 0.0285), BAL TNF-α (E; *P = 0.0146), IL-6 (F; *P = 0.0070), and CXCL-1 (G; *P = 0.0087). In addition, septic mice pretreated with liraglutide tended toward lower BAL IL-10 (I), wet-to-dry lung weight ratios (J), and BAL RAGE content (K). BAL IFN-γ (H) and protein (L) as well as sepsis-induced percent change in body weight (M) and bacterial burden in the spleen (N) were not different between liraglutide and caloric restriction. Finally, BAL cell counts were significantly higher in liraglutide-pretreated mice compared with caloric restriction (D; 0.0052). n = 10–17 [Statistical analysis: two-way ANOVA (mixed model) + Sidak’s multiple comparisons test (A); log-rank (Mantel-Cox; B); Mann–Whitney U test (C, J–N); two-tailed unpaired t test on natural log-transformed data (D–I)]. Each data point represents an individual animal. Data in A were summarized as median ± 95% confidence interval. Data in B were summarized as survival portions. Data in C–N were summarized as boxplots, where the box encompassed the interquartile range, error bars encompassed minimum to maximum values, and horizontal line showed median. BAL, bronchoalveolar lavage; CXCL-1, C-X-C motif chemokine ligand 1; IFN-γ, interferon gamma; IL-1β, interleukin-1β; IL-6, interleukin-6; IL-10, interleukin-10; RAGE, receptor for advanced glycation end products; TNF-α, tumor necrosis factor-α.