Effects of Post‐Resuscitation Normoxic Therapy on Oxygen‐Sensitive Oxidative Stress in a Rat Model of Cardiac Arrest

Abstract

Background

Cardiac arrest (CA) can induce oxidative stress after resuscitation, which causes cellular and organ damage. We hypothesized that post‐resuscitation normoxic therapy would protect organs against oxidative stress and improve oxygen metabolism and survival. We tested the oxygen‐sensitive reactive oxygen species from mitochondria to determine the association with hyperoxia‐induced oxidative stress.

Methods and Results

Sprague–Dawley rats were subjected to 10‐minute asphyxia‐induced CA with a fraction of inspired O₂ of 0.3 or 1.0 (normoxia versus hyperoxia, respectively) after resuscitation. The survival rate at 48 hours was higher in the normoxia group than in the hyperoxia group (77% versus 28%, P<0.01), and normoxia gave a lower neurological deficit score (359±140 versus 452±85, P<0.05) and wet to dry weight ratio (4.6±0.4 versus 5.6±0.5, P<0.01). Oxidative stress was correlated with increased oxygen levels: normoxia resulted in a significant decrease in oxidative stress across multiple organs and lower oxygen consumption resulting in normalized respiratory quotient (0.81±0.05 versus 0.58±0.03, P<0.01). After CA, mitochondrial reactive oxygen species increased by ≈2‐fold under hyperoxia. Heme oxygenase expression was also oxygen‐sensitive, but it was paradoxically low in the lung after CA. In contrast, the HMGB‐1 (high mobility group box‐1) protein was not oxygen‐sensitive and was induced by CA.

Conclusions

Post‐resuscitation normoxic therapy attenuated the oxidative stress in multiple organs and improved post‐CA organ injury, oxygen metabolism, and survival. Additionally, post‐CA hyperoxia increased the mitochondrial reactive oxygen species and activated the antioxidation system.

Nonstandard Abbreviations and Acronyms

AGE

advanced glycation end product

CA

cardiac arrest

HMGB

high mobility group box

HO

heme oxygenase

ROS

reactive oxygen species

RQ

respiratory quotient

VCO₂

carbon dioxide production

VO₂

oxygen consumption

Clinical Perspective

What Is New?

• The results reported here indicate that cardiac arrest is associated with mitochondrial dysfunction and that the amount of mitochondrial reactive oxygen species generation could be doubled in the brain when these are exposed to hyperoxic conditions.

What Are the Clinical Implications?

• This study provides much‐needed evidence of the augmentation of the O₂‐sensitivity of mitochondria in the generation of reactive oxygen species after cardiac arrest.

• Mitochondrial reactive oxygen species generation during hyperoxia is therefore likely a causal factor in upregulating the oxidizing pathways.

• Our finding implies the important pathophysiology linked with post‐cardiac arrest organ failure.

Cardiac arrest (CA) is a major public health issue affecting ≈600 000 people each year in the United States. ¹ New therapies to reduce the occurrence and extent of organ injury, and the resulting pathophysiology, are imperative to improve patient survival and quality of life after CA. ² , ³

In the context of reperfusion injury, hyperoxia is thought to cause cellular damage by increasing the generation of reactive oxygen species (ROS), resulting in an exacerbation of oxidative toxic stress, deterioration of mitochondrial dysfunction, and derangement of cellular metabolism. ⁴ , ⁵ , ⁶ Compelling data from clinical and preclinical trials have demonstrated benefits of post‐CA normoxic therapy. ⁴ , ¹⁷

The purpose of post‐resuscitation normoxic therapy is to reduce ROS generation and consequently attenuate oxidative stress after CA. Although the rationale for post‐resuscitation normoxic therapy is well established, ³ , ⁷ , ¹⁰ , ¹¹ its mechanism and specifically the role of mitochondrial ROS in post‐CA pathophysiology have not been clearly elucidated. Evidence has shown that mitochondria play a crucial role as effectors and targets of ischemia/reperfusion injury such as CA. ¹⁸ , ¹⁹ , ²⁰ , ²¹ Mitochondrial ROS generation is theoretically oxygen‐sensitive ²² ; indeed, decreased production of ROS by isolated mitochondria was observed when the O₂ was lowered below that of the air‐saturated medium. ²³ However, though it is widely accepted that CA victims are afflicted by hyperoxia‐induced organ damage ¹² caused by an exacerbation of oxidative stress attributable to ongoing mitochondrial dysfunction, ²⁴ the O₂‐sensitive mitochondrial ROS generation and its augmentation by CA has not been shown.

We previously reported on a system‐level derangement of post‐CA metabolism characterized by dissociated oxygen consumption (VO₂) and carbon dioxide production (VCO₂), resulting in a lowered respiratory quotient (RQ). This newly found phenotype was O₂ sensitive. ²⁵ In the present study, we use a rat cardiopulmonary resuscitation (CPR) model to test (1) whether hyperoxia‐induced injury is mediated by mitochondrial ROS production and increased inflammation and (2) whether normoxic post‐resuscitation therapy protects against oxidative stress in several organs and improves systematic oxygen metabolism.

METHODS

The Institutional Animal Care and Use Committees of the Feinstein Institutes for Medical Research approved this study protocol. We performed all instrumentation and surgical preparation according to our previously described protocol. ²⁵ Details are found in Data S1. The data that support the findings of this study are available from the corresponding author upon reasonable request.

Comparing Hyperoxia and Normoxia

Sprague–Dawley male rats (n=30) were randomly assigned into 2 groups 10 minutes after CPR. Successfully resuscitated animals were given normoxic (30% oxygen, inhaled, n=15) or hyperoxic (100% oxygen, inhaled, n=15) therapy (Figure S1A). Oxygen therapy continued for up to 2 hours and monitoring for up to 48 hours, at which point neurological deficit score was taken as described previously. ²⁶

VO₂, VCO₂, and RQ Measurements

Using the mechanical ventilation circuit, we evaluated the system‐level metabolic alteration for up to 120 minutes via VO₂, VCO₂, and RQ (RQ=VCO₂/VO₂ as previously described ²⁵ ). We added 2 major modifications to obtain VO₂ more accurately at a fraction of inspired O₂ of 1.0. First, we used a CO₂ mainstream capnometer (Nihon Kohden Corp., Tokyo, Japan) to avoid sampling (suctioning) the gas from the ventilation system. Secondly, we measured the molecular ratio of inhalation to exhalation, which was an independent measurement from the gas concentration measurements. The values were calculated and reported as standard temperature and pressure.

Immunochemical Assays of Oxidative Stress Indicators, HO‐1, and HMGB‐1

Carbonyl protein and 8‐hydroxy‐2′‐deoxyguanosine levels were measured as an indicator of oxidative stress. Heme oxygenase (HO)‐1 is a cytoprotective antioxidant enzyme, ²⁷ and the HO‐1 level was measured as an indicator of the activation of the antioxidant system. The HMGB‐1 (high mobility group box‐1) protein is an inflammatory alarmin that is released following non‐programmed cell death, but by apoptotic cells. ²⁸ We measured HMGB‐1 levels as an indicator of the activation of inflammatory pathways. These biomarkers were measured by ELISA based on the commercial protocol.

Isolation of Brain and Kidney Mitochondria and Evaluation of Mitochondrial Respiratory Function

Mitochondrial samples were collected from the sham‐normoxia and CA‐normoxia groups. All isolations were performed at 4°C. Brain and kidney mitochondria were isolated using a procedure modified from Kim et al. ²⁰ , ²⁹ The oxygen consumption was measured using a Strathkelvin oxygen electrode (30°C). ADP‐dependent (state 3) and ADP‐limited (state 4) respirations were measured in 150 µL of the mitochondrial suspension (0.5 mg/mL) using glutamate and malate as substrates.

Comparing Mitochondrial H₂O₂ Generation in Ex Vivo Normoxic and Hyperoxic Conditions

H₂O₂ generation in mitochondria isolated from brain and kidney was used as a measure of mitochondrial ROS and compared between sham‐normoxia and CA‐normoxia experimental groups. Two different O₂ tension settings were applied to the mitochondria of each group. Different O₂ concentrations in the medium were achieved by mixing nitrogen‐ and air‐saturated buffers (Figure S2). H₂O₂ levels were determined using an Amplex Red hydrogen peroxide/peroxidase assay kit (Invitrogen, Carlsbad, CA, USA), as instructed by the manufacturer. The mitochondria were incubated at 0.025 mg of protein/mL at 30°C. H₂O₂ production was initiated in the mitochondria using glutamate (10 mmol/L) and malate (2.5 mmol/L) as substrates, using an established protocol ³⁰ (Figure S3).

Wet/Dry Weight Ratio of the Lung

The right lower lobe from each animal was weighed immediately after collection and then placed into a 60°C oven to dry. After 3 days, the tissue was weighed to determine the wet‐to‐dry lung weight ratio (W/D).

Immunofluorescence Staining of HO‐1 and HMGB‐1 and Histological Lung Injury Evaluation

Double immunostaining was performed for HMGB‐1 or advanced glycation end product (AGE) in combination with HO‐1. Stained sections were observed under an LSM 880 confocal imaging system (Carl Zeiss, Inc., Jena, Germany) and a BZ‐X800 all‐in‐one fluorescence microscope (Keyence, Elmwood Park, NJ, USA). We analyzed the data using the BZ‐X800 analyzer software (Keyence, Elmwood Park, NJ, USA). The lung sections were also stained with hematoxylin and eosin. A blinded investigator reviewed the histopathology using a modified acute lung injury scoring system, as previously described by Kawamura et al. ³¹

Statistical Analysis

Data are shown as the means and SD for continuous variables and the counts and frequencies for categorical variables. Mann‐Whitney U test was used to compare 2 independent groups for continuous variables. For multi group comparisons, Kruskal‒Wallis test with the Dunn‐Bonferroni approach were used. Survival rates were estimated by the Kaplan‒Meier method, and the Wilcoxon test was used to compare the groups. Two‐tailed P values were calculated, and P<0.05 was considered statistically significant. SPSS 25.0 (IBM, Armonk, NY, USA), JMP 10.1 (SAS Institute, Cary, NC, USA), and GraphPad Prism 8 (GraphPad Software, San Diego, CA, USA) were used for statistical analyses.

RESULTS

Post‐Resuscitation Normoxic Therapy Improved Neurological Function and Survival After CA

The normoxic therapy group demonstrated a higher survival rate (77%) at 48 hours after resuscitation compared with the hyperoxia group (28%, P=0.010: Figure S1B). Along with improved survival rates, the normoxic therapy group had significantly lower neurological deficit score (359±140) compared with the hyperoxia group (452±85, P=0.026: Figure S1C).

Post‐Resuscitation Normoxic Therapy Reduced System‐Level Dissociations of Oxygen Metabolism

At 120 minutes after CPR, the sham group had normal values of VO₂, VCO₂, and RQ (14.3±1.5 mL/kg per minute, 14.1±1.4 mL/kg per minute, and 0.99±0.12, respectively). At 120 minutes after CPR, the VO₂ in the CA‐normoxia group was significantly lower over time (17.8±3.1 mL/kg per minute) than in the CA‐hyperoxia group (31.1±5.2 mL/kg per minute, Kruskal‒Wallis; P=0.002, pairwise P=0.003: Figure 1B). There were no differences in VCO₂ between the groups (Kruskal‒Wallis; P=0.080: Figure 1C). As a result, the RQ after CA was significantly higher in the CA‐normoxia group (0.81±0.05) than in the CA‐hyperoxia group (0.58±0.03, Kruskal‒Wallis; P=0.001, pairwise P=0.001: Figure 1D), and the value for the normoxia group improved to the generally cited normal range of 0.7 to 1.0.

Figure 1. Post‐resuscitation normoxic therapy reduced system‐level dissociations of oxygen metabolism in a rat cardiopulmonary resuscitation model.

A, PaO₂ levels over time compared among the groups. There were no significant differences in the values of PaO₂ at 10 minutes between the normoxia and hyperoxia (302±76 and 361±71 Torr, respectively). In the normoxic therapy group, PaO₂ was successfully maintained at 120±10 Torr during the initial 120 minutes after randomization. PaO₂ levels in the hyperoxia group were higher than 350 Torr at all times. Time (min) from starting CPR is depicted on the x‐axis. B, VO₂ over time. Values were averaged every 5 minutes. The VO₂ in the CA‐normoxia group was significantly lower over time. C, Carbon dioxide production over time. Values were averaged every 5 minutes. There were no differences in carbon dioxide production between the groups. D, Respiratory quotient over time. The respiratory quotient after CA was significantly higher in the CA‐normoxia group than it was in the CA‐hyperoxia group, and the value for the normoxia group improved to the generally cited normal range of 0.7 to 1.0; mean±SD, Kruskal‒Wallis test with the post hoc analysis. CA indicates cardiac arrest; CPR, cardiopulmonary resuscitation; STP, standard temperature and pressure.

Post‐Resuscitation Normoxic Therapy Protected Organs Against Oxidative Stress

During hyperoxia, an increase in the amounts of carbonyl protein in the brain were observed in the CA group (sham‐normoxia and hyperoxia, 0.15±0.14 and 0.33±0.05 nmol/mg protein; CA‐normoxia and hyperoxia, 0.32±0.23 and 0.93±0.13 nmol/mg protein; Kruskal‒Wallis; P<0.001, pairwise P=0.396, P=0.005, respectively: Figure 4A). The carbonyl protein levels were O₂‐sensitive, and a similar trend was observed in the sham group. However, hyperoxia‐induced a more substantial increase in carbonyl protein levels in the CA group.

Figure 4. Hyperoxia increased heme oxygenase‐1 (HO‐1) levels and cardiac arrest (CA) increased both HO‐1 and HMGB‐1 (high mobility group box 1) levels in the brain.

A, Immunohistochemistry studies of the choroid plexus and subependymal cell layer areas in sham animals. B, The choroid plexus and subependymal cell layer areas in CA animals. C, The hippocampus area in sham animals. D, The hippocampus area in CA animals. Images were captured in a low power field and the yellow bars represent a scale of 50 µm. E, Ratio of HO‐1 to 4′,6‐diamidino‐2‐phenylindole fluorescent stain in the hippocampus area. F, HO‐1 levels in the brain by ELISA. Hyperoxia induced an increase in HO‐1 expression in both the sham and CA groups. Based on a comparison between the sham and CA animals at normoxia (P<0.01), CA also seemed to induce HO‐1 expression. G, Ratio of HMGB‐1 to 4′,6‐diamidino‐2‐phenylindole fluorescent stain in the hippocampus area. H, HMGB‐1 levels in the brain by ELISA. CA increased the HMGB‐1 levels. There may be a correlation between O₂ dependency and HMGB‐1 levels, but this was only observed post‐CA. ^# P<0.05, ^## P<0.01 vs sham normoxia. CA indicates cardiac arrest; HMGB‐1, high mobility group box 1; HO‐1, heme oxygenase 1; DAPI is a 4′,6‐diamidino‐2‐phenylindole fluorescent stain. *P<0.05, **P<0.01 vs sham hyperoxia. ^$ P<0.05, ^$$ P<0.01 vs CA normoxia; mean±SD, Kruskal‒Wallis test with the post hoc analysis.

Additionally, hyperoxia induced an increase in carbonyl protein in the lung in both the sham and CA groups (sham‐normoxia and hyperoxia, 0.13±0.07 and 0.42±0.07 nmol/mg protein; CA‐normoxia and hyperoxia, 0.28±0.15 and 1.00±0.14 nmol/mg protein; Kruskal‒Wallis; P<0.001, pairwise P=0.028, P=0.006, respectively: Figure 2B). Hyperoxia also increased 8‐hydroxy‐2′‐deoxyguanosine levels in the tracheal secretion in the CA group (sham‐normoxia and hyperoxia, 3.2±1.5 and 5.6±0.5 ng/mL; CA‐normoxia and hyperoxia, 7.6±1.2 and 20±6 ng/mL; Kruskal‒Wallis; P<0.001, pairwise P=0.901, P=0.528, respectively: Figure 2D). CA‐induced increase in 8‐hydroxy‐2′‐deoxyguanosine levels was greater than in the sham animals in both normoxia and hyperoxia settings (P=0.006 and P=0.002, respectively).

Figure 2. Post‐resuscitation normoxic therapy reduced carbonyl protein levels in the organs and 8‐hydroxy‐2′‐deoxyguanosine (8OHdG) levels in tracheal secretions and urine.

A, Carbonyl protein levels in the brain. B, Carbonyl protein levels in the lung. C, Carbonyl protein levels in the kidney. The carbonyl protein levels were O₂ dependent, and there was a trend in O₂ dependency in the sham group. However, hyperoxia‐induced increases in carbonyl protein were more remarkable in the cardiac arrest (CA) group (P<0.01). D, 8OHdG levels in tracheal secretions. Hyperoxia also increased the 8OHdG levels of the tracheal secretion in the CA group. A supply of 100% oxygen via the endotracheal tube did not affect the 8OHdG levels in the tracheal secretion samples from sham animals. Consequently, O₂ dependency was only observed in the CA animals (P<0.01). E, 8OHdG levels in urine samples. A supply of 100% oxygen did not affect the 8OHdG levels in the urine samples from sham animals. Consequently, O₂ dependency was only observed in CA animals (P<0.01). CA indicates cardiac arrest; and 8OHdG, 8‐hydroxy‐2′‐deoxyguanosine. ^## P<0.01 vs sham normoxia. **P<0.01 vs sham hyperoxia. ^$$ P<0.01 vs CA normoxia; mean±SD, Kruskal‒Wallis test with the post hoc analysis.

A hyperoxia‐induced increase in carbonyl protein in the kidney was also observed in the CA group (sham‐normoxia and hyperoxia, 0.08±0.03 and 0.12±0.04 nmol/mg protein; CA‐normoxia and hyperoxia, 0.44±0.18 and 0.99±0.37 nmol/mg protein; Kruskal‒Wallis; P<0.001, pairwise P=1.000, P=0.814, respectively: Figure 2C). Hyperoxia also increased 8‐hydroxy‐2′‐deoxyguanosine levels in the urine of the CA group (sham‐normoxia and hyperoxia, 87±63 and 39±36 ng/mL; CA‐normoxia and hyperoxia, 65±34 and 168±81 ng/mL; Kruskal‒Wallis; P=0.005, pairwise P=0.659, P=0.099, respectively: Figure 2E).

These results support the concept that oxidative stress is O₂‐sensitive overall and can be seen in multiple organs even in non‐injured animals. However, there is an augmentation of oxygen‐sensitivity induced by CA. Thus, post‐resuscitation normoxic therapy significantly protected the animals against oxidative toxic stress.

Cardiac Arrest Induces Mitochondrial Respiratory Dysfunction and the Augmentation of O₂‐Sensitive Mitochondrial H₂O₂ Generation

Oxidative Phosphorylation

The state 3 respiration activity of the brain and kidney mitochondria in the CA‐normoxia group (209±26 and 148±37 nmol/min per mg) declined significantly compared with those of the sham‐normoxia group (286±50 and 269±55 nmol/min per mg; P=0.003, P=<0.001, respectively: Figure 3A and 3B). In contrast, the state 4 respiration activity of the brain and kidney mitochondria did not change. As a result, the respiratory control ratio showed an altered trend after CA in both tissues.

Figure 3. Cardiac arrest (CA) decreased state 3 mitochondrial respiratory activity and increased the degree of hyperoxia‐induced increases in mitochondrial H₂O₂ generation.

A, Mitochondrial respiratory activity in the brain. B, Mitochondrial respiratory activity in the kidney. The state 3 respiration activity of the brain and the kidney mitochondria in the CA‐normoxia group declined significantly compared with that of the sham‐normoxia group. In contrast, the state 4 respiration activity of the brain and the kidney mitochondria did not change significantly. ^&& P<0.01 vs sham. C, Mitochondrial H₂O₂ generation in the brain. D, Mitochondrial H₂O₂ generation in the kidney. The ex vivo hyperoxic condition significantly accelerated the H₂O₂ generation of the brain and the kidney mitochondria in both the sham and CA groups. E, O₂ dependency of mitochondrial H₂O₂ generation compared between CA and sham animals. The brain mitochondria generated approximately twice as much H₂O₂ under hyperoxia. The kidney mitochondria also showed a similar trend, but there was no statistical significance. CA indicates cardiac arrest. ^# P<0.05, ^## P<0.01 vs sham normoxia. ^$ P<0.05, ^$$ P<0.01 vs CA normoxia. ^& P<0.05 vs sham; mean±SD, Mann‐Whitney U test.

H₂O₂ Generation in Mitochondria

Ex vivo hyperoxia significantly accelerated H₂O₂ generation in brain mitochondria in both the sham and CA groups (sham‐normoxia and hyperoxia, 66±13 and 84±29 pmol/mg protein; CA‐normoxia and hyperoxia, 55±26 and 115±23 pmol/mg protein; Kruskal‒Wallis; P=0.002, pairwise P=1.000, P=0.001, respectively: Figure 3C). A similar trend was also observed in kidney mitochondria (sham‐normoxia and hyperoxia, 45±15 and 61±16 pmol/mg protein; CA‐normoxia and hyperoxia, 44±14 and 89±39 pmol/mg protein; Kruskal‒Wallis; P=0.009, pairwise P=0.733, P=0.020, respectively: Figure 3D).

The effect of ex vivo hyperoxia on mitochondrial H₂O₂ generation was calculated for each mitochondrial sample, and the numbers were averaged and compared between the groups. The brain mitochondria generated approximately twice as much H₂O₂ in the hyperoxic condition (sham and CA, 125±20% and 267±203%; P=0.036, respectively: Figure 3E). The kidney mitochondria also showed a similar trend, but there was no statistical significance (sham and CA, 146±43% and 268±250%; P=0.279, respectively: Figure 3E). These results support the augmentation of O₂‐sensitive mitochondrial H₂O₂ generation after CA. This result might be attributed to the mitochondrial dysfunction observed in the CA rat model.

Hyperoxia Increased HO‐1 Levels and Cardiac Arrest Increased Both HO‐1 and HMGB‐1 Levels in the Brain

Immunohistochemical studies of the brain revealed an O₂‐sensitive increase in HO‐1 expression (Figure 4A through 4D).The ratio of HO‐1 positive to 4′,6‐diamidino‐2‐phenylindole fluorescent stain positive areas (Figure  4E) showed a similar trend with the quantitative measurement of HO‐1 levels by ELISA (sham‐normoxia and hyperoxia, 0.39±0.03, 0.95±0.18 ng/mg protein; CA‐normoxia and hyperoxia, 0.77±0.13 and 1.32±0.17 ng/mg protein, Kruskal‒Wallis; P<0.001, pairwise P=0.010, P=0.017, respectively: Figure 4F).

Immunohistochemical studies of the brain also revealed a CA‐induced increase in HMGB‐1 (Figure 4A through 4D). The ratio of HMGB‐1 positive to 4′,6‐diamidino‐2‐phenylindole fluorescent stain‐positive areas (Figure 4G) showed a similar trend with the quantitative measurement of HMGB‐1 levels (sham‐normoxia and hyperoxia, 38.1±5.5 and 42.8±3.1 ng/mg protein; CA‐normoxia and hyperoxia, 125±4 and 133±2 ng/mg protein, Kruskal‒Wallis; P<0.001, pairwise P=1.000, P=0.659, respectively: Figure 4H). The O₂‐sensitive augmentation was only observed post‐CA.

Cardiac Arrest was Associated With Hyperoxic Lung Injury

The lung W/D ratio revealed that hyperoxia increased lung edema after CA (sham‐normoxia and hyperoxia, 4.4±0.6 and 4.7±0.2; CA‐normoxia and hyperoxia, 4.6±0.4 and 5.6±0.5, Kruskal‒Wallis; P=0.006, pairwise P=1.000, P=0.033, respectively: Figure S4A). The lung injury score histologically similarly exhibited increased the degree of lung injury at 120 minutes after CPR because of hyperoxia (sham‐normoxia and hyperoxia, 2.7±2.3 and 2.3±1.0; CA‐normoxia and hyperoxia, 4.3±2.9 and 14±2; Kruskal‒Wallis; P=0.002, pairwise P=1.000, P=0.157, respectively: Figure S4B and S4C). These results support the idea that hyperoxia induces lung injury, which is more substantial post‐CA.

Hyperoxia Increased HO‐1 Levels and Cardiac Arrest Increased Both HO‐1 and HMGB‐1 Levels in the Lung

Immunohistochemical studies of the lung revealed an O₂‐sensitive increase in HO‐1 expression in the sham animals (Figure 5A through 5D). However, this increase was suppressed by hyperoxia after CA. The ratio of HO‐1‐positive to 4′,6‐diamidino‐2‐phenylindole fluorescent stain‐positive (Figure 5E) areas showed that hyperoxia increased HO‐1 in the sham group but not in the CA group, and quantitative ELISA showed a similar trend to the immunohistochemistry results (sham‐normoxia and hyperoxia, 5.2±0.5 and 12.4±1.0 ng/mg protein; CA‐normoxia and hyperoxia, 10.5±1.4 and 6.5±0.5 ng/mg protein; Kruskal‒Wallis; P<0.001, pairwise P<0.01, P=0.274, respectively: Figure 5F).

Figure 5. Hyperoxia increased heme oxygenase‐1 (HO‐1) levels and cardiac arrest (CA) increased both HO‐1 and HMGB‐1 levels in the lung.

A, Immunohistochemistry studies of the lung in sham animals. B, Immunohistochemistry studies in the CA animals. Images are shown as high‐power field and the white bars represent a scale of 10 µm. C, Immunohistochemical studies in the sham animals. D, Immunohistochemical studies in the CA animals. Immunohistochemical studies of the lung indicated an increase in AGE expression, especially in the CA‐hyperoxia group. Images are shown as low‐power field and the yellow bars represent a scale of 50 µm. E, Ratio of HO‐1 to 4′,6‐diamidino‐2‐phenylindole fluorescent stain in the Lung. F, HO‐1 levels in the lung by ELISA. Hyperoxia increased the HO‐1 in the sham group but not in the CA group. There was an increase and suppression of HO‐1 by the hyperoxia after CA. G, Ratio of HMGB‐1 to 4′,6‐diamidino‐2‐phenylindole fluorescent in the lung. H, HMGB‐1 Levels in the lung by ELISA. CA but not hyperoxia increased the HMGB‐1 levels in the lung. There might be a correlation between O₂ dependency and HMGB‐1 levels, but it was only observed post‐CA. AGEs indicates advanced glycation end products; CA, cardiac arrest; DAPI, 4′,6‐diamidino‐2‐phenylindole fluorescent stain; HMGB‐1, high mobility group box 1; and HO‐1, heme oxygenase 1. ^# P<0.01, ^## P<0.05 vs sham normoxia. **P<0.01 vs sham hyperoxia. ^$ P<0.05, ^$$ P<0.01 vs CA normoxia; mean±SD, Kruskal‒Wallis test with the post hoc analysis.

Immunohistochemical studies also revealed an increase in HMGB‐1 and AGE in the lung after CA (Figure 5A through 5D). Quantitative measurement of HMGB‐1 by ELISA (Figure 5G) showed a trend similar to the immunohistochemistry results (sham‐normoxia and hyperoxia, 199±46, 221±36 ng/mg protein; CA‐normoxia and hyperoxia, 471±51 and 716±49 ng/mg protein; Kruskal‒Wallis; P<0.001, pairwise P=1.000, P=0.528, respectively: Figure 5H). These results suggest that HO‐1 expression in the lung is O₂‐sensitive, but there may also be a paradoxical suppression of HO‐1 expression post‐CA during hyperoxia. CA increased HMGB‐1 levels in the lung. Given the immunohistochemical finding of AGE expression in conjunction with an increase in HMGB‐1, our results suggest that the inflammatory HMGB‐1 pathway may be activated post‐CA.

DISCUSSION

Our data demonstrate that oxidative stress is O₂‐sensitive overall and occurs in multiple organs, but the increase in oxidative stress is more notable after CA. The pathophysiological response to elevated supplemental O₂ varied between post‐CA animals and normal animals because of the oxidative stress. It can be inferred that an increase in oxidants (harmful pathways) and/or down‐regulation of antioxidant enzymes (protective pathways) may occur after resuscitation with hyperoxia. The post‐resuscitation normoxic therapy regimen significantly protected the animals against oxidative stress and further improved their survival after CA. The results reported here indicate that CA is associated with mitochondrial dysfunction and that the amount of mitochondrial H₂O₂ generation could be doubled in the brain and kidney when these are exposed to hyperoxic conditions. Thus, this study provides much‐needed evidence of the augmentation of the O₂‐sensitivity of mitochondria in the generation of ROS after CA. Mitochondrial ROS generation during hyperoxia is therefore likely a causal factor in upregulating the oxidizing pathways.

Previously, exciting data from our laboratory revealed new information about post‐resuscitation metabolic phenotypes: we showed a system‐level derangement of metabolism characterized by the O₂‐sensitive dissociation of oxygen consumption and carbon dioxide production, resulting in a lowered RQ. ²⁵ Studies have shown that VO₂ is not dependent on O₂ levels in healthy animals and humans. ³² , ³³ This finding is consistent with our biochemical understanding of mitochondrial respiration, which suggests that the electron transport chain is limited by nicotinamide adenine dinucleotide. ³⁴ , ³⁵ The relevance of this to our findings is the prolonged production of ROS within the electron transport chain enzymes. Under normal electron transfer, 4 electrons are required to reduce molecular oxygen to 2 water molecules. However, the creation of superoxide requires a single electron to reduce molecular oxygen to a single molecule of superoxide, ³⁶ in which there is a shift in stoichiometry from normal oxygen usage to a requirement for more oxygen.

Heme oxygenase (HO‐1) is a rate‐limiting enzyme that catalyzes the oxidation of heme to biologically active molecules: iron, a gene regulator; biliverdin, an antioxidant; and carbon monoxide, a heme ligand. ³⁷ HO‐1 expression can confer cytoprotection in many lung ³⁸ and vascular disease models. ³⁹ , ⁴⁰ In line with the findings of others, ³⁷ there was an increase in HO‐1 expression by hyperoxia in our study, suggesting that oxidative stress might upregulate the cytoprotective pathways within the enzymatic antioxidant role of HO‐1. In addition, the HO‐1 levels were high post‐CA, reflecting the demands of activating protective pathways because of the ongoing post‐CA pathophysiology. ⁴¹ In this scenario, hyperoxia seems to accelerate the expression of HO‐1 because of the exacerbated oxidative stress after CA. However, the lung with hyperoxia showed a paradoxically low expression of HO‐1, which was observed in conjunction with an increase in the inflammatory HMGB‐1 and, subsequently, the receptor for AGE (RAGE) pathway. The cross‐reaction between HO‐1 and the HMGB‐RAGE pathway regulated by peroxisome proliferator‐activated receptors was suggested in a previous report using acute lung injury models. ⁴² Our immunohistochemical study revealed expression of AGE ligands, which contribute to a variety of microvascular and macrovascular complications by binding to RAGE. ⁴³ , ⁴⁴ Our results suggest that the HMGB‐1/RAGE pathway was activated in the lung, particularly when the tissues were exposed to high supplemental oxygen after CA/reperfusion, and this overexpression of the inflammatory pathway may account for the paradoxical suppression of HO‐1 during hyperoxia after CA. Other possible explanations for the difference in HO‐1 expression in the lungs compared with other organs may be differences in the time course of organ responses. It is plausible that other organs may have suppressed HO‐1 levels in later, unobserved periods post‐resuscitation, but the lungs showed it much earlier and within the timeframe we observed.

This study is subject to several limitations. First, the use of a rat model to study post‐CA metabolism has inherent limitations in representing human disease. ⁴⁵ Secondly, our study does not distinguish the contribution of mitochondrial ROS generation from other sources of ROS, and there are multiple non‐mitochondrial enzymes associated with ROS generation, such as Nicotinamide adenine dinucleotide phosphate oxidase, xanthine oxidase, and monoamine oxidase. ²⁴ These enzymes are possible contributors to the oxidative stress that we observed in the CA animals, and these reactions are theoretically O₂‐sensitive. However, our study was not designed to test for non‐mitochondrial enzymes. Further investigation may focus on non‐mitochondrial ROS generation and may warrant a systematic approach to evaluate the contribution of mitochondrial function to the overall generation of ROS. Thirdly, the use of isolated mitochondria is valuable for studying their contribution in oxygen‐mediated ROS generation post‐CA but does sacrifice some physiological context of the host tissues, necessitating a more tissue‐based or in vivo approach. Finally, our study was not designed to reveal cause of death in our asphyxia‐CA rat model. In nature, CA is defined as a sudden cessation of blood flow and, when the heart successfully regains the circulation, the body suffers from ischemia/reperfusion injury in multiple organ systems, and neurological function is generally the most affected. ⁴⁶ , ⁴⁷ CA may or may not be attributed to a cardiac etiology, although cardiac origin is the most prevalent in humans. ⁴⁸ , ⁴⁹ , ⁵⁰ Our study protocol of 10 minutes asphyxia may not be severe enough to affect heart function, despite the damage observed in other organs. No differences or pattern alterations were observed in Mean arterial pressure, heart rate, or blood lactate levels (representing systemic circulatory dysfunction) between the hyperoxia and normoxia groups (Figure S5A through S5C). Hyperoxia‐induced augmentation of oxidative stress may worsen neurological function; however, we did not observe any augmentation in cardiac or circulatory function in the early period. Moreover, in our experimental animal model, we did not identify an increase of oxidative stress in the heart (Figure S6). Our data do not directly point to a cause of death of our animal model, but previous studies that investigated outcomes in a rat model of asphyxia‐CA showed pathological and functional deficits in neurons after resuscitation, ²⁶ and this neuronal damage may be worsened by hyperoxia. ⁴ , ¹⁷ Therefore, it is plausible that the cause of death of our asphyxia‐CA model is linked to devastating neurological deficits rather than the circulatory function, once the animal has regained spontaneous circulation. However, our experimental setting did not have continuous Electroencephalography or video monitoring to allow us to assess this. Future work will focus on the cause of death by closely monitoring neurological function.

CONCLUSIONS

Our post‐resuscitation normoxic therapy reduced oxidative stress in multiple organs and improved organ damage, oxygen metabolism, and survival after CA. The mitochondrial ROS generation was O₂‐sensitive, and mitochondrial generation of ROS was increased after CA. HO‐1 expression was also O₂‐sensitive. However, a paradoxical suppression of HO‐1 was observed in the lung with a concomitant upregulation of HMGB‐1.

Sources of Funding

None.

Disclosures

K. Saeki is an employee of Nihon Kohden Corporation and Nihon Kohden Innovation Center, INC. There are no products in market to declare. This does not alter the authors' adherence to all the journal's policies on sharing data and materials. Shinozaki and Becker have a patent right of metabolic measurements in critically ill patients. Shinozaki has a grant/research support from Nihon Kohden Corp. Becker has a grant/research support from Philips Healthcare, the NIH, Nihon Kohden Corp., Zoll Medical Corp, PCORI, BrainCool, and United Therapeutics and owes patents including 7 issued patents and several pending patents involving the use of medical slurries as human coolant devices to create slurries, reperfusion cocktails, and measurement of respiratory quotient. The remaining authors have no disclosures to report.

Supporting information

Data S1

Figures S1–S6

References 51 , 52

(J Am Heart Assoc. 2021;10:e018773. DOI: 10.1161/JAHA.120.018773.)