Potential conflict of interest: Nothing to report.
Obstructive sleep apnea (OSA) is characterized by chronic intermittent hypoxia (CIH). OSA is associated with nonalcoholic steatohepatitis (NASH) in obese subjects. The aim of this study was to investigate the effects of CIH on the liver in the absence of obesity. Lean C57BL/6J mice (n = 15) on a regular chow diet were exposed to CIH for 12 weeks and compared with pair-fed mice exposed to intermittent air (IA, n = 15). CIH caused liver injury with an increase in serum ALT (224 ± 39 U/l versus 118 ± 22 U/l in the IA group, P < 0.05), whereas AST and alkaline phosphatase were unchanged. CIH also induced hyperglycemia, a decrease in fasting serum insulin levels, and mild elevation of fasting serum total cholesterol and triglycerides (TG). Liver TG content was unchanged, whereas cholesterol content was decreased. Histology showed swelling of hepatocytes, no evidence of hepatic steatosis, and marked accumulation of glycogen in hepatocytes. CIH led to lipid peroxidation of liver tissue with a malondialdehyde (MDA)/free fatty acids (FFA) ratio of 0.54 ± 0.07 mmol/mol versus 0.30 ± 0.01 mmol/mol in control animals (P < 0.01), and increased levels of active nuclear factor kappaB (NF-κB) in the nuclear fraction of hepatocytes, suggesting that CIH induced oxidative stress in the liver. Finally, CIH greatly exacerbated acetaminophen-induced liver toxicity, causing fulminant hepatocellular injury. Conclusion: In the absence of obesity, CIH leads to mild liver injury via oxidative stress and excessive glycogen accumulation in hepatocytes and sensitizes the liver to a second insult, whereas NASH does not develop. (HEPATOLOGY 2007;45:1007–1013.)
Obstructive sleep apnea (OSA) is characterized by recurrent collapse of the upper airway during sleep, leading to chronic intermittent hypoxia (CIH).1 OSA is a common disease, present in 2% of women and 4% of men in the general U.S. population; however, the prevalence rises to 40% to 60% in obese individuals.2, 3 CIH of OSA has been associated with an increased risk of hypertension, type 2 diabetes, dyslipidemia, and atherosclerosis, independent of underlying obesity.2, 4–9 Moreover, studies in rodent models of intermittent hypoxia (IH) demonstrated that CIH can cause hypertension,10 insulin resistance,11 and dyslipidemia.12, 13 Thus, CIH of OSA has been implicated in causality of cardiovascular and metabolic disorders, independent of obesity.
An emerging body of evidence indicates that OSA is associated with non-alcoholic steatohepatitis (NASH) and chronic liver injury in obese individuals.14, 15 Whether OSA can confer risk of NASH, independent of obesity, remains unclear. Two major mechanisms have been implicated in NASH: (1) hepatic steatosis, which is linked to insulin resistance; and (2) increased levels of oxidative stress with liver injury and subsequent inflammation.16, 17 We previously showed that CIH leads to progression of hepatic steatosis and insulin resistance in leptin-deficient obese mice.12 However, the effects of CIH on hepatic lipids in the absence of obesity have not been examined. Whereas OSA and CIH induce oxidative stress and inflammation in multiple organs and tissues,18–21 the impact of CIH on levels of lipid peroxidation and inflammation in the liver has not been studied. We hypothesized that oxidative stress of CIH will independently lead to liver injury, which will be further exacerbated by another hepatic insult.
The purpose of the current study was to examine the effects of CIH on the liver in the absence of obesity. We exposed C57BL/6J mice to CIH for 12 weeks and examined: (1) serum activity of liver enzymes as well as fasting serum insulin and blood glucose levels; (2) lipid content, lipid peroxidation, levels of inflammatory cytokines, and activity of nuclear factor kappa B (NF-κB) in the liver; (3) liver histopathology; (4) an interaction between CIH and a hepatic prooxidant acetaminophen on the liver.
A total of 38 wild-type, 6- to 8-week-old male, lean C57BL/6J mice purchased from The Jackson Laboratory (Bar Harbor, ME) were used in the study. The study was approved by the Johns Hopkins University Animal Care and Use Committee and complied with the American Physiological Society Guidelines for Animal Studies. For all blood samples, injections, and surgical procedures, anesthesia was induced and maintained with 1% to 2% isoflurane administered through a facemask.
A gas control delivery system was designed to regulate the flow of room air, nitrogen, and oxygen into customized cages housing the mice as previously described.11 A maximum of three mice were housed continuously in a single customized cage (dimensions 27 × 17 × 17 cm) with constant access to food and water. During each period of IH, the forced inspiratory oxygen was reduced from 20.9% to 4.9% ± 0.1% over a 30-second period and then rapidly reoxygenated to room air levels in the subsequent 30-second period.
Fifteen mice were placed in the IH chamber for 12 consecutive weeks. Food intake and body weight were monitored daily for each animal. All animals were kept in a controlled environment (22–24°C with a 12-hour:12-hour light:dark cycle; lights on at 9:00 AM) on a standard Purina chow diet (4% fat) with free access to water. Fifteen mice were exposed to intermittent room air (IA, control group) for 5 days in identical chambers and were weight-matched to the CIH group daily during the experiment by varying food intake (Table 1). Weight-matching was conducted in pairs. The CIH and IA states were induced during the light phase, alternating with 12 hours of constant room air during the dark phase.
Table 1. Baseline Characteristics of Lean C57BL/6J Mice Exposed to Intermittent Hypoxia for 12 Weeks
In a separate experiment, animals were exposed to either CIH (n = 4) or IA control conditions (n = 4) for 12 weeks as described. At the end of the exposure, mice were injected intraperitoneally with acetaminophen (Sigma-Aldrich Corp., St. Louis, MO) at 600 mg/kg body weight and killed 24 hours later.
Mice fasted for 5 hours before bleeding and killing. Arterial blood (1 ml) was obtained by direct cardiac puncture under 1% to 2% isoflurane anesthesia. Serum was separated and frozen at −80°C. After blood withdrawal, the animals were killed with pentobarbital (60 mg, intraperitoneally). Liver was surgically removed and weighed. Liver tissue was either frozen in liquid nitrogen and kept at −80°C for further biochemical studies, or fixed in buffered 10% formalin for histological examination, or fixed in buffered 4% glutaraldehyde for electron microscopy, or immediately frozen in Sakura Tissue-Tek ornithine carbamoyl-transferase Compound (Sakura Finetek USA Inc., Torrance, CA)
Serum ALT, AST, and alkaline phosphatase activity, serum total bilirubin, and albumin levels were measured by the Clinical Chemistry Laboratory of the Johns Hopkins Bayview Medical Center. Fasting serum cholesterol, phospholipids, free fatty acids (FFA), and triglycerides were measured using test kits from Wako Diagnostics Inc. (Richmond, VA). Serum leptin and insulin levels were detected with enzyme-linked immunosorbent assay (ELISA) kits from Linco Research Inc. (St. Charles, MO). Serum corticosterone was determined by ELISA Kit from R&D Systems Inc. (Minneapolis, MN). Fasting blood glucose was measured with Accu-Chek Comfort Curve kit from Roche Diagnostics Inc. (Indianapolis, IN). Liver was homogenized using Omni EZ Connect Homogenizer (Omni International, Warrington, VA). Lipids were extracted from the liver with chloroform-methanol, according to Bligh-Dyer procedure,22 and total cholesterol, phospholipid, FFA, and triglyceride content was measured with kits from Wako Diagnostics. Malondialdehyde (MDA) was measured in total liver lysate with a kit from Oxford Biomedical Research Inc. (Oxford, MI). Lipid peroxidation in liver tissue was assessed by determining a ratio of MDA content, which is a product of polyunsaturated fatty acid peroxidation, to FFA content. For measurements of a transcription factor NF-κB, a nuclear fraction was isolated as described.13 Activated NF-κB was measured in the nuclear fraction using an ELISA kit from Oxford Biomedical Research. Interleukin-1β (IL-1β), tumor necrosis factor 1α, and macrophage-inflammatory protein-2 levels in total liver lysate were determined with ELISA kits from R&D Systems, Inc. (Minneapolis, MN). IL-6 levels in the liver were measured with an ELISA kit from RayBiotech, Inc. (Norcross,GA), and myeloperoxidase levels in the liver were measured with an ELISA kit from HyCult Biotechnology (Uden, The Netherlands). Glycogen liver content was determined using modified phenol-sulfuric acid technique as described.23
Paraffin-embedded tissue was sectioned at 3- 5-μm thickness and stained with hematoxylin-eosin (HE) or periodic acid-Schiff (PAS). Ornithine carbamoyl-transferase frozen tissue was sectioned at 7- to 10-μm thickness and stained with oil red O. Affinity-purified rabbit anti-human/mouse caspase 3 active antibodies (R&D Systems Inc., Minneapolis, MN) and the terminal deoxynucleotidyl transferase-mediated nick-end labeling assay by Roche GmbH (Germany) were used to detect apoptosis. Histological assessment of tissue morphology was performed using an Olympus light microscope (Olympus, Tokyo, Japan) and evaluated in a blinded fashion (by M.S.T.).
Liver tissues were fixed with a solution containing 2.5% glutaraldehyde, 2% paraformaldehyde in 0.1 M sodium phosphate buffer (pH 7.4) at 4°C. The specimens were postfixed in 2% osmium tetroxide solution, rinsed in distilled water. After dehydration and embedding in epoxy resin, semithin sections were stained with toluidine blue for orientation. Ultrathin sections were routinely processed and examined with electron microscope (Carl Zeiss Inc., Germany).
All values are reported as mean ± SEM. Comparisons within and between the groups of mice were performed using general linear model analysis of variance (ANOVA), paired t test (between day 0 and day 5 within a group) or unpaired t test (between IH and IA groups). A P value of less than 0.05 was considered significant.
Baseline Characteristics, Serum and Liver Biochemistry in C57BL/6J Mice Exposed to CIH.
Mice exposed to CIH were weight-matched throughout the exposure to the control animals (Table 1). Despite reduced food intake in both groups in comparison with day 0, all mice gained 1 to 2 g weight by the end of the experiment. CIH did not have an effect on liver weight. A significant CIH-induced increase in fasting blood glucose was seen, in combination with a decrease in fasting serum insulin levels, which suggests that CIH may lead to insulin deficiency. CIH doubled circulating levels of leptin, which was consistent with our previous observations in short-term IH.11, 13, 24 Serum corticosterone levels were identical in both groups (Table 1), suggesting that mice exposed to CIH did not exhibit evidence of stress. CIH resulted in an increase in fasting serum total cholesterol and triglyceride levels (Table 2), which was similar to our previous observations.13, 24 Liver cholesterol content was significantly depleted during the exposure, whereas liver triglyceride, phospholipid, and FFA content remained unchanged (Table 2), revealing a marked difference with ob/ob mice, which had significant increases in liver triglyceride and phospholipid content after 12 weeks' exposure to CIH.12
Table 2. Fasting Serum Lipids and Liver Lipid and Cytokine Content in Lean C57BL/6J Mice Exposed to Intermittent Hypoxia for 12 Weeks
Chronic Intermittent Hypoxia
Abbreviations: IL, interleukin; MIP-2, macrophage-inflammatory protein-2.
Liver injury in CIH was assessed by measuring of the levels of key liver enzymes in the serum. CIH resulted in a 2-fold increase in activity of serum ALT, whereas activity of AST and alkaline phosphatase was unchanged (Fig. 1). CIH had no effect on serum total bilirubin or albumin levels (Table 1).
Sections of liver showed no significant inflammation or fibrosis in either treated animals or controls (Fig. 2A,B). No evidence was seen for cholestasis or bile duct injury, and no macrovesicular steatosis was seen. The control animals revealed very mild (grade 1+ on a scale of 0–3) microvesicular steatosis, which was evident in sections stained with HE (Fig. 2A, arrows) and confirmed by staining with oil red O (Fig. 2C, arrows). The CIH animals, in contrast, showed no microvesicular steatosis but exhibited moderate hepatocellular swelling with rarefaction of the cytoplasm (Fig. 2B, arrows), whereas lipid droplets essentially disappeared (Fig. 2D). Biochemical measurements showed a decrease in liver cholesterol content in mice exposed to CIH, whereas liver triglyceride and phospholipids content remained unchanged (Table 2). PAS stain showed a marked increase in the liver glycogen content in CIH animals (Fig. 3A, B), which was confirmed by EM (Fig. 3C,D). The biochemical assay demonstrated that the amount of glycogen in the liver almost tripled after CIH in comparison with control conditions (Fig. 3E). There was no change in a number of apoptotic hepatocytes at hypoxic conditions according to the terminal deoxynucleotidyl transferase-mediated nick-end labeling assay and caspase-3 immunohistochemistry (not shown).
Lipid Peroxidation and NFκB Activity in the Liver.
CIH induced lipid peroxidation in hepatic tissue (Fig. 4A), suggesting that CIH led to oxidative stress in the liver. Hypoxic animals also exhibited a greater than threefold increase in levels of active isoform of NF-κB in the nuclear fraction of total liver lysate (Fig. 4B). CIH did not increase hepatic levels of myeloperoxidase and cytokines IL-1β, IL-6, macrophage-inflammatory protein-2 (Table 2), and tumor necrosis factor alpha levels were undetectable, suggesting that NF-κB activation by IH did not result in the significant inflammatory response in the liver.
Effect of CIH on Acetaminophen-Induced Liver Toxicity.
Acetaminophen was injected at 600 mg/kg 24 hours before completion of 12 weeks' exposure to CIH or IA control conditions. In control mice, acetaminophen caused a 2-fold increase in serum ALT in comparison with baseline (Figs. 1 and 5A). Three of 4 injected control mice showed normal liver histopathology (Fig. 5B), and 1 animal exhibited necrotic changes involving 10% of zone 3. CIH led to marked exacerbation of acetaminophen toxicity with a greater than 10-fold increase in serum ALT and AST (Fig. 5A), and necrosis in liver tissue in all animals, involving 50% to 100% of zone 3 (Fig. 5C).
The purpose of the study was to determine whether CIH, similar to that observed in human OSA, causes liver injury and steatohepatitis in lean C57BL/6J mice. Several major findings resulted from the study. First, CIH caused elevation of serum ALT levels in association with increased lipid peroxidation in the liver and active NF-κB in the nuclear fraction of the liver tissue, as well as histological evidence of hepatocyte swelling, which suggests that CIH leads to liver injury via oxidative stress mechanisms. Second, CIH did not increase liver lipid content, which indicates that CIH does not cause hepatic steatosis and steatohepatitis in the absence of obesity. Third, CIH caused hyperglycemia in association with low serum insulin levels and glycogen accumulation in the liver, which suggests that CIH may cause glycogenic hepatopathy. Fourth, CIH greatly increased acetaminophen hepatotoxicity, suggesting that CIH may predispose to fulminant hepatocellular injury. In the discussion that follows, we explore the relationships between CIH, the mechanisms of liver injury, and hepatic metabolic pathways and discuss clinical implications of our work.
CIH, Liver Injury, and Oxidative Stress.
Severe OSA is associated with elevation in liver enzymes and biopsy evidence of liver injury.14, 25 Moreover, the severity of liver injury in patients with OSA directly correlates with the severity of the hypoxic insult.26 Our findings provide direct experimental evidence that CIH causes liver injury. Deleterious effects of CIH can be attributed to the intermittent nature of the stimulus, because sustained hypoxemia (e.g., high altitudes) rarely leads to pathological changes in the liver.27, 28 Indeed, the liver normally functions at relatively low levels of oxygen tension, because three-quarters of the liver blood supply is provided by portal circulation.29, 30 Chronic sustained hypoxemia may affect liver function, as in the case of severe right-sided heart failure resulting in ischemia and congestive hepatomegaly.31–33 In our experiments, no evidence was seen of hepatic congestion after exposure to CIH (Fig. 2, Table 1) and no other evidence of right-sided heart failure. Acute hypoxemia also may lead to acute liver injury when the insult is very severe, for example, shock with profound tissue ischemia.34, 35 Our experimental animals appeared to thrive and gained weight during 12 weeks of CIH. Thus, the liver injury during CIH is not likely to be induced by either heart failure with subsequent congestive hepatomegaly or by severe acute hypoxia per se.
CIH results in repetitive cycles of hypoxia and reoxygenation, leading to excessive production of reactive oxygen species and oxidative stress in various organs and tissues.36 OSA is associated with increased serum levels of MDA and 8-isoprostane, which are products of lipid peroxidation.18, 36 Experimentally induced CIH in rodents causes lipid peroxidation in different organs. CIH in mice increases MDA and isoprostane levels in the brain as well as activity of NADPH oxidase, an enzyme-producing superoxide.19, 20 CIH in rats increases MDA levels in the myocardium and decreases activity of superoxide dismutase, an important endogenous antioxidant.21 Oxidative stress has been implicated in pathogenesis of systemic complications of CIH, including neurological impairments and atherosclerosis.37–39 We provide the first direct experimental evidence that CIH, similar to that observed in severe OSA,40 induces a nearly 2-fold increase in lipid peroxidation in the liver (Fig. 4A), which may cause impairment of the cellular membranes and lead to liver injury.16 Oxidative stress is a likely mechanism of NF-κB activation in liver tissue,41 which may play a protective role, inhibiting apoptosis42 and suppressing generation of reactive oxygen species.41 Interestingly, NF-κB activation by CIH did not produce a significant inflammatory response in the liver (Table 2).
Our data indicate that the independent effect of CIH on the liver was relatively mild. In contrast, interaction of CIH and acetaminophen greatly augmented both insults. Acetaminophen is a hepatic prooxidant that depletes glutathione.43 Therefore, synergy between acetaminophen and CIH could be a result of diminished antioxidant defense. Thus, CIH may predispose to fulminant hepatocellular injury in the presence of the second insult.
CIH and NASH.
Obese patients with OSA exhibit both components of NASH, hepatic steatosis and inflammation.14 We have previously shown that CIH for 12 weeks augments pre-existing hepatic steatosis in ob/ob mice.12 OSA is also independently associated with insulin resistance and type 2 diabetes mellitus in obese individuals,2, 44–46 and CIH exacerbates insulin resistance in obese rodents.11 Because hepatic steatosis is a known consequence of insulin resistance,16, 47, 48 a CIH-induced increase in insulin resistance in combination with oxidative stress may lead to NASH in obese patients with OSA and in obese mice exposed to CIH. However, our new findings indicate that, in the absence of obesity, CIH does not result in hyperinsulinemia and liver steatosis (Tables 1 and 2, Fig. 2C,D). Based on our previously published12 and current data, we propose that a combination of obesity and CIH may be necessary for the development and progression of NASH.
CIH and Glycogenic Hepatopathy.
An unexpected finding of our study was a nearly threefold increase in liver glycogen content in mice exposed to CIH (Fig. 3). The histology of the liver in mice exposed to CIH resembles human glycogenic hepatopathy, which exhibits the following features: (1) glycogen accumulation in hepatocytes, which appears as hepatocyte swelling on HE stain (Fig. 2A, B) and abundant intracellular deposition of PAS-positive material on the PAS stain (Fig. 3A, B); (2) no evidence of fatty liver; (3) minimal inflammation, no or minimal lobular necrosis and fibrosis.49 Glycogenic hepatopathy may lead to an increase in serum transaminase levels49 as we observed in mice exposed to CIH.
The cause of glycogenic hepatopathy in mice exposed to CIH is not clear. Glycogen hepatopathy has been described in patients with diabetes mellitus type 1 and poor glucose control.49, 50 High concentrations of glucose may inactivate glycogen phosphorylase and induce glycogen synthase, causing glycogen accumulation in hepatocytes.50 Our current data demonstrate that CIH in lean mice leads to hyperglycemia in association with a decrease in serum insulin level, which suggests insulin deficiency. The persistent hyperglycemia in mice subjected to CIH may have induced glycogen hepatopathy.
Conclusions and Clinical Implications.
Our data indicate that CIH causes liver injury, oxidative stress, and excessive glycogen accumulation in the livers of lean mice, whereas neither uncomplicated steatosis nor NASH develop. The effects of CIH were greatly exacerbated by exposure to acetaminophen. We speculate that CIH may cause liver injury in OSA, especially in patients concomitantly exposed to another hepatic insult.