Ethyl pyruvate attenuates ventilation‐induced diaphragm dysfunction through high‐mobility group box‐1 in a murine endotoxaemia model

Abstract Mechanical ventilation (MV) can save the lives of patients with sepsis. However, MV in both animal and human studies has resulted in ventilator‐induced diaphragm dysfunction (VIDD). Sepsis may promote skeletal muscle atrophy in critically ill patients. Elevated high‐mobility group box‐1 (HMGB1) levels are associated with patients requiring long‐term MV. Ethyl pyruvate (EP) has been demonstrated to lengthen survival in patients with severe sepsis. We hypothesized that the administration of HMGB1 inhibitor EP or anti‐HMGB1 antibody could attenuate sepsis‐exacerbated VIDD by repressing HMGB1 signalling. Male C57BL/6 mice with or without endotoxaemia were exposed to MV (10 mL/kg) for 8 hours after administrating either 100 mg/kg of EP or 100 mg/kg of anti‐HMGB1 antibody. Mice exposed to MV with endotoxaemia experienced augmented VIDD, as indicated by elevated proteolytic, apoptotic and autophagic parameters. Additionally, disarrayed myofibrils and disrupted mitochondrial ultrastructures, as well as increased HMGB1 mRNA and protein expression, and plasminogen activator inhibitor‐1 protein, oxidative stress, autophagosomes and myonuclear apoptosis were also observed. However, MV suppressed mitochondrial cytochrome C and diaphragm contractility in mice with endotoxaemia (P < 0.05). These deleterious effects were alleviated by pharmacologic inhibition with EP or anti‐HMGB1 antibody (P < 0.05). Our data suggest that EP attenuates endotoxin‐enhanced VIDD by inhibiting HMGB1 signalling pathway.

process that involves oxidative loads, muscle atrophy (arising from calpain, caspase-3, the autophagy-lysosomal pathway [ALP] and ubiquitin-proteasome system [UPS] activation), structural damage and muscle fibre remodelling, have not been fully explored. [4][5][6] Therefore, a detailed understanding of the precise mechanisms underlying VIDD is critical for the development of potential strategies to reduce the prolonged use of MV and thus ICU mortality.
Sepsis is a major risk factor for ICU patients developing diaphragm dysfunction. 2,3,7 During the progress of sepsis, deleterious host response to infectious constituents, such as lipopolysaccharide (LPS), may induce an inflammatory cascade and cause ventilator-associated pneumonia (VAP), subsequently leading to multiple organ failure. 5, [8][9][10] Animal studies have demonstrated that infection is a principal cause of abnormal diaphragm activities. 8,11 Sepsis-mediated diaphragmatic weakness and VIDD involve common molecular pathways, namely excessive oxidative loads and mitochondrial abnormalities within the injured diaphragm myofibrils, suggesting that sepsis may be an accessory contributor to VIDD. 5, 6,8,10 In an acute lung injury (ALI), reactive oxygen species (ROS) are the major oxidants in the diaphragm and can be produced in mitochondria, sarcoplasmic reticula, sarcolemma, transverse tubes and cytosol within 6 hours of MV. 12,13 Furthermore, sepsis and MV-induced oxidative stress may deteriorate diaphragm contractility and are critical contributors to proteolytic pathway activation. [14][15][16] The primary proteases in the skeletal muscle consist of (1) UPS, (2) calcium-related proteases and (3) lysosomal enzymes. 10,17 The up-regulation of muscle-specific ubiquitin E3 ligases F-box protein atrogin-1 and muscle RING-finger proteins-1 (MuRF-1) is crucial for the proteolysis of monomeric myofibrillar proteins in the diaphragms of animals and patients using MV. [16][17][18] Increased autophagosome formation also occurs in MV-augmented diaphragmatic weakness, as reflected in an elevation of autophagic biomarker microtubule-related protein light chain (LC) 3. [19][20][21] Mitochondria are a principal source of diaphragmatic ROS and act as a pivotal upstream regulator that induces the molecular pathways engendering diaphragm muscle atrophy during endotoxaemia or MV. 22,23 In addition, myonuclear apoptosis can also be accelerated by mitochondrial ROS, elevated cellular calcium levels and sarcoplasmic reticulum stress-induced activation of calpain and caspase-3. 20,24 Furthermore, sepsis and MV-induced oxidative stress may up-regulate the production of inflammatory mediators, including high-mobility group box 1 (HMGB1), interleukin 6 (IL- 6) and plasminogen activator inhibitor-1 (PAI-1), 9,15,[25][26][27][28][29] and subsequently impair diaphragm activities.
Rodent studies of endotoxaemia have revealed that toll-like receptor 4 (TLR4) modulates diaphragm injury by activating the nuclear factor-κB (NF-κB) pathway. 11,30 In our previous mouse study investigating sepsis, the reduced ALP and mitigated mitochondrial ultrastructural changes were observed by inhibiting TLR4/NF-κB signalling through genetic manipulation of TLR4 using homozygous knockout. 31 TLR4 is the most renowned receptor of the TLR family and is crucial for the recognition of the damage-associated molecular pattern (DAMP), including extracellular matrix components, HMGB1 and LPS. 30,32 Although inhibition of TLR4 could be an effective therapeutic strategy for VIDD, in light of the potential risk of increased infectious complications using this approach, targeting endogenous TLR4 ligands such as HMGB1 could be more prudent. Related studies have demonstrated that the expression levels of HMGB1 are associated with diaphragmatic dysfunction in caecal ligation and puncture in animals and that the administration of anti-HMGB1 antibodies can effectively attenuate sepsis-induced diaphragm dysfunction in septic rats. 26,28 Several related studies have demonstrated that ethyl pyruvate, a potent free radical scavenger, can inhibit HMGB1 production and improve the survival of critically ill patients. 33,34 Nevertheless, the use of ethyl pyruvate for the management of VIDD is still unexplored.
Murine endotoxaemia models have been employed to recapitulate human sepsis for nearly a century. 35 In this study, we investigated that the impact of MV with or without LPS and related HMGB1 signalling contributes to VIDD using a murine model of endotoxaemia. We hypothesized that the administration of ethyl pyruvate or anti-HMGB1 antibody would diminish HMGB1 expression, diaphragmatic structural damage, generation of free radicals, proteolytic protein synthesis and mitochondrial dysfunction and restore muscle contractility in the diaphragm of mice with or without endotoxaemia exposed to MV.

| Experimental groups
Animals were randomly distributed into six groups in each experiment: group 1, non-ventilated control wild-type mice with normal saline; group 2, non-ventilated control wild-type mice with LPS; group 3, tidal volume (V T ) 10 mL/kg wild-type mice with normal saline; group 4, V T 10 mL/kg wild-type mice with LPS; group 5, V T 10 mL/kg wild-type mice after ethyl pyruvate (100 mg/kg) administration with LPS; group 6, V T 10 mL/kg wild-type mice after anti-HMGB1 antibody (100 mg/kg) with LPS; group 7, V T 10 mL/kg wild-type mice after ethyl pyruvate (50 mg/kg) administration with LPS; group 8, V T 10 mL/kg wild-type mice after anti-HMGB1 antibody (50 mg/kg) administration with LPS; group 9, V T 10 mL/kg wildtype mice after anti-HMGB1 isotype control antibody (100 mg/kg) administration with LPS. In groups 1-6, three mice underwent transmission electron microscopy (TEM) and specific force, and five mice underwent measurement for immunohistochemistry assay, inflammatory cytokines, protein carbonyl groups, superoxide dismutase, terminal deoxynucleotidyl transferase-mediated dUTP-biotin nick end-labelling (TUNEL) assay, and Western blots. In groups 7-9, five mice underwent measurement for HMGB1 and Western blots.

| Ventilator protocol
We used our established murine model of VILI, as described previously. 9,31 Briefly, a 20-gauge angiocatheter was inserted into the tracheotomy orifice of mice and general anaesthesia was maintained by regular intraperitoneal administration of zoletil 50 (5 mg/kg) and xylazine (5 mg/kg) at the beginning of the experiment and every Schellekens et al and our previous work demonstrated that 8-hour MV in mice can induce increased cytokine expression, lysosomal autophagy and diaphragm atrophy. 9,31 At the end of the study, heparinized blood was taken from the arterial line for analysis of arterial blood gas, and the mice were sacrificed. The non-ventilated control mice were anaesthetized and sacrificed immediately.

| LPS and pharmacological inhibitors
Mice were administrated intravenously with either 1 mg/kg of Salmonella typhosa LPS (Lot 81H4018; Sigma Chemical Co., St. Louis, MO) or an equivalent volume of normal saline via the internal jugular vein as a control. After 1 hour of spontaneous respiration to allow for developing a septic response, the mice were subjected to MV for 8 hours. 9,31,36 Two doses of ethyl pyruvate (Sigma, St Louis, MO) were administered intraperitoneally. The first dose was administered 30 minutes before the mice were subjected to MV and the second dose was used after the mice were subjected to 4 hours of MV. 29,34 Anti-HMGB1 antibody 100 mg/kg (chicken anti-pig HMGB1 polyclonal antibody, SHINO-TEST, Tokyo, Japan) or isotype control antibody (non-immune immunoglobulin G, SHINO-TEST, Tokyo, Japan) was administered intravenously 30 minutes before the start of MV. 28 The doses of ethyl pyruvate and anti-HMGB1 were chosen on the basis of our and other studies that showed 100 mg/kg ethyl pyruvate or anti-HMGB1 had better effects on inhibiting HMGB1 activity. 28,29,34 2.5 | Detection of cytokines in the bronchoalveolar lavage fluid PAI-1, with a lower detection limit of 0.02 ng/mL, and HMGB1 (1 ng/ mL) were detected in bronchoalveolar lavage (BAL) fluid using immunoassay kits containing primary polyclonal anti-mouse antibodies that were cross-reactive with rat and mouse PAI-1 and HMGB1 (PAI-1: Molecular Innovations, Inc, Southfield, MI; HMGB1: Shino-Test corporation, Kanagawa, Japan). Each sample was run in duplicate, according to the manufacturer's instructions.

| Immunohistochemistry
The diaphragms were paraffin embedded, sliced at 4 μm, deparaffinized, antigen unmasked in 10 mmol/L sodium citrate (pH 6.0), incubated with rabbit HMGB1 primary antibody (1:100; Santa Cruz Biotechnology, Santa Cruz, CA) and biotinylated goat anti-rabbit secondary antibody (1:100) according to the manufacturer's instruction for an immunohistochemical kit (Santa Cruz Biotechnology, Santa Cruz, CA). The specimens were further conjugated with horseradish peroxidase-streptoavidin complex, detected with a diaminobenzidine (DAB) substrate mixture and counterstained by haematoxylin. A dark-brown DAB signal, identified by arrows, indicated positive staining of HMGB1 of muscle fibres, whereas shades of light blue signified non-reactive cells.

| Real-time polymerase chain reaction
For isolating total RNA, the lung tissues were homogenized in TRIzol reagents (Invitrogen Corporation, Carlsbad, CA), according to the manufacturer's instructions. Total RNA (1 μg) was reverse transcribed using a GeneAmp polymerase chain reaction (PCR) system 9600 (PerkinElmer, Life Sciences, Inc, Boston, MA), as previously described. 31 The following primers were used for real-time polymerase chain reaction: HMGB1, forward primer 5′-TGGCAAAGGCTGACAAGGCTC-3′ and reverse primer 5′-GGATGCTCGCCTTTGATTTTGG-3′ and GAPDH as internal control, forward primer 5′-GGAGCGAGACCCCACTAACA-3′ and reverse primer 5′-ACATACTCAGCACCGGCCTC-3′ (Protech Technology Enterprise Co., Ltd., Taipei, Taiwan). 37,38 All quantitative PCR reactions using SYBR Master Mix were performed on a CFX96 Touch Real-Time PCR Detection system (Bio-Rad Laboratories, Inc, Hercules, CA). All PCR reactions were performed in duplicate and heated to 95°C for 5 minutes followed by 40 cycles of denaturation at 95°C for 10 seconds, and annealing at 55°C for 30 seconds. The relative gene expression was calculated using 2 −ΔΔCT method and the standard curves (cycle threshold values vs template concentration) were prepared for each target gene and for the internal control (GAPDH) in each sample. The specific gene's cycle threshold (Ct) values were normalized to the GAPDH and compared with the nonventilated control group with LPS that was assigned a value of 1 to calculate the relative fold change in expression.

| Statistical evaluation
The Western blots were quantitated using a National Institutes of Health (NIH) image analyser Image J 1.27z (National Institutes of F I G U R E 1 Electron microscopy and muscle force-frequency activity of the diaphragm. Representative micrographs of the longitudinal sections of diaphragm (×20 000: upper panel; ×40 000: lower panel) were from the same diaphragms of non-ventilated control mice and mice ventilated at a tidal volume (V T ) of 10 mL/kg (V T 10) for 8 h with or without LPS administration (n = 3 per group). (A and B) Non-ventilated control wild-type mice with or without LPS treatment: normal sarcomeres with distinct A bands, I bands and Z bands; (C) 10 mL/kg wild-type mice without LPS treatment (normal saline): increase of diaphragmatic disarray; (D) 10 mL/kg wild-type mice with LPS treatment: disruption of sarcomeric structure with loss of mitochondrial swelling, streaming of Z bands and collection of lipid droplets; (E) 10 mL/kg wild-type mice pretreated with ethyl pyruvate: reduction of diaphragmatic disruption. (F) Diaphragm musclespecific force production was measured as described in Materials and Methods2. Mitochondrial swelling with concurrent loss of cristae and autophagosomes containing heterogeneous cargo are identified by arrows. Ethyl pyruvate, 100 mg/kg, was given intraperitoneally 30 min before mechanical ventilation and 4 h after mechanical ventilation. *P < 0.05 vs the non-ventilated control mice with LPS treatment; † P < 0.05 vs all other groups. Scale bar represents 500 nm. EP, ethyl pyruvate; Hz, hertz; LPS, lipopolysaccharide; N, Newton Health, Bethesda, MD) and presented as arbitrary units. Values were expressed as the mean ± SD from at least five separate experiments. The data of protein oxidation, superoxide dismutase, specific force, histopathologic assay and oxygenation were analysed using Statview 5.0 (Abascus Concepts, Cary, NC; SAS Institute). All results of real-time PCR and Western blots were normalized to the non-ventilated control wild-type mice with LPS. ANOVA was used to assess the statistical significance of the differences, followed by multiple comparisons with a Scheffe′s test, and a P < 0.05 was considered statistically significant. We have performed the Shapiro-Wilk normality test and verify that all data are parametric (P > 0.05).
Additional details, including measurement of diaphragm force-frequency relationships, immunoblot analysis, immunohistochemistry, mitochondrial isolation, measurement of diaphragmatic oxidative stress and antioxidant enzyme expression, TEM and TUNEL assay were performed as previously described. 9,31 3 | RE SULTS

| Reduction of the effects of MV on endotoxinenhanced VIDD, oxygen radicals and inflammatory cytokines using ethyl pyruvate
We employed MV (10 mL/kg) at room temperature for 8 hours to elicit VIDD in mice. The physiological conditions at the beginning and end of MV are listed in Table S1. Normovolemic status was sus- were from the non-ventilated control mice and mice ventilated at a tidal volume of 10 mL/kg for 8 h with or without LPS administration (n = 5 per group). Western blots were performed using antibodies that recognize calpain (E), atrogin-1 (F), MuRF-1 (G) and GAPDH expression from the diaphragms of non-ventilated control mice and mice ventilated at a tidal volume of 10 mL/ kg for 8 h with or without LPS administration (n = 5 per group). Arbitrary units were expressed as relative calpain, atrogin-1 and MuRF-1 activation (n = 5 per group). Ethyl pyruvate, 100 mg/kg, was given intraperitoneally 30 min before mechanical ventilation and 4 h after mechanical ventilation. *P < 0.05 vs the non-ventilated control mice with LPS treatment; † P < 0.05 vs all other groups. BAL, bronchoalveolar lavage; GAPDH, glyceraldehydes-phosphate dehydrogenase; HMGB1, high-mobility group box-1; MuRF-1, muscle ring finger-1; PAI-1, plasminogen activator inhibitor-1; SOD, sodium dismutase in this study were described in supplementary data ( Figure S1). TEM was performed to explore MV-and LPS-induced changes in the dia- was observed in mice with endotoxaemia subjected to V T 10 mL/kg compared to those without endotoxaemia subjected to V T 10 mL/kg and the non-ventilated control mice. The elevated HMGB1 expression levels after MV were substantially diminished after the administration of either ethyl pyruvate or anti-HMGB1 antibody ( Figure 3B).

| Reduction of the effects of MV on endotoxinaugmented VIDD by ethyl pyruvate and anti-HMGB1 antibody
To

| Suppression of the effects of MV on endotoxin-enhanced expression of caspase-3 and diaphragm apoptosis by ethyl pyruvate and anti-HMGB1 antibody
Studies have demonstrated that caspase-3 is crucial to the intrinsic apoptotic pathway. 20    were from the non-ventilated control mice and mice ventilated at a tidal volume of 10 mL/kg for 8 h with or without LPS administration (n = 5 per group). Western blots were performed using antibodies that recognize calpain (E), atrogin-1 (F), MuRF-1 (G) and GAPDH expression from the diaphragms of non-ventilated control mice and mice ventilated at a tidal volume of 10 mL/kg for 8 h with or without LPS administration (n = 5 per group). Arbitrary units were expressed as relative calpain, atrogin-1 and MuRF-1 activation (n = 5 per group). Anti-HMGB1 antibody, 100 mg/kg, was administered intravenously 30 min before the start of ventilation. *P < 0.05 vs the non-ventilated control mice with LPS treatment; † P < 0.05 vs all other groups autophagy. These deleterious effects were attenuated by using the anti-HMGB1 antibody. Although ROS and early cytokines are important to contribute to VIDD, and herein, we provide the evidence that using anti-HMGB1 antibody can elevate the SOD activity and suppress the oxidants, inflammatory cytokines and VIDD compara- decreased NF-κB and plasma cytokines in 3 h, but those inflammatory parameters were elevated in 9 hours in high-dose LPS (30 mg/ kg)-challenged mice for survival test. 55 In the subsequent study, ethyl pyruvate at different doses of 100 mg/kg or 50 mg/kg was proven to reduce the mortality from endotoxin-induced ALI and the permeability index in mice. 34 Notably, the beneficial effects of 100 mg/kg ethyl pyruvate were superior compared to those of 50 mg/kg, which are consistent with our dose-response results ( Figure S1). A single intraperitoneal dose of ethyl pyruvate was able to reduce methaemoglobinaemia induced by dapsone for a short period of time of 3-6 h. 56 Thus, a second dose was designed to be given after the mice subjected to 4 hours of MV at our protocol. As for the toxicity of ethyl pyruvate, dosage up to 150 mg/kg was proven to be a safe and effective medication in endotoxaemic horses, and ethyl pyruvate (90 mg/kg) at 6-hour intervals for five more doses was used with safety in human trials. 57,58 In addition, anti-HMGB1 antibody (100 mg/kg) can significant attenuate diaphragm dysfunction in septic rats, and the beneficial effect was comparable to the highest dose of 250 mg/kg, but was superior compared to the dose of 25 mg/kg, which are in accordance with our results ( Figure S1). 28 The present study had some limitations. First, in activated neutrophils, HMGB1 induced p38 MAPK, extracellular signal-regulated kinases 1/2 and serine/threonine kinase/protein kinase B. It also increased the production of NF-κB-dependent inflammatory cytokines, including tumour necrosis factor-α, IL-6 and macrophage inflammatory protein-2, in lung tissue during infection and injury. 37,42 Evidence has indicated that acetylation of HMGB1 prevents its entry into the nucleus and causes the release of HMGB1 from cells, thus initiating inflammation, as well as LPS-activated acetylation of HMGB1, through Janus Kinase/STAT signalling in TLR4-activated macrophages. 27,59 Further investigation is necessary to explore other possible molecular mechanisms that involve HMGB1 and the advantageous effects of ethyl pyruvate in preventing patients with endotoxaemia from developing VIDD. Second, we utilized the anti-HMGB1 antibody in the study since HMGB1-/-mice die shortly after birth due to a defect in the transcriptional activation of the glucocorticoid receptor. 29 However, Kim et al reported that intranasal delivery of HMGB1 siRNA provides the neuroprotection in the ischaemic brain mediated by target gene knockdown. 60  Academic Editing, for their help with the experiment.

CO N FLI C T O F I NTE R E S T
The authors confirm that there are no conflicts of interest. F I G U R E 7 Schematic figure illustrating the signalling pathway activation with mechanical ventilation and endotoxaemia. Endotoxinmediated augmentation of mechanical stretch-induced inflammatory cytokine production and diaphragm injury was alleviated after the administration of ethyl pyruvate and anti-HMGB1 antibody. 31 HMGB1, high-mobility group box-1; LC3-II, light chain 3-II; LPS, lipopolysaccharide; MuRF-1, muscle ring finger-1; NF-κB, nuclear factor κappa B; PAI-1, plasminogen activator inhibitor-1; ROS, reactive oxygen species; TLR4, toll-like receptor 4; VIDD, ventilator-induced diaphragm dysfunction. *Reference 31