Abstract
It is unknown whether nicotinamide adenine dinucleotide phosphate (NADPH) oxidase 2 (Nox2) activation is early associated with endotoxemia and liver damage in nonalcoholic fatty liver disease (NAFLD). To address this issue, we evaluated Nox2 activation, oxidative stress, gut permeability, and lipopolysaccharide (LPS) serum levels in 67 children with biopsy-proven NAFLD and 73 controls. Compared with controls, NAFLD patients had higher Nox2 activity, isoprostane, zonulin, and LPS levels. Multivariate linear regression analysis showed that triglycerides, high-density lipoprotein (HDL), homeostatic model assessment-estimated insulin resistance (HOMA-IR), LPS, and isoprostanes were independently associated with Nox2-derivative peptide (sNox2-dp) levels. Within the NAFLD group, patients with nonalcoholic steatohepatitis (NASH) had significant higher levels of sNox2-dp, isoprostanes, LPS, triglycerides, HOMA-IR, fasting glucose and insulin, and lower HDL than those without NASH. Furthermore, sNox2-dp levels were linearly associated with the histological grading of steatosis, inflammation, ballooning, fibrosis, and NAFLD activity score. This study provides evidence that children with NAFLD have Nox2 overactivation compared with controls and significant association with the degree of liver damage. The close relationship between Nox2 and LPS serum levels leads to hypothesize a potential role for gut-derived LPS in eliciting systemic Nox2 activation.
Introduction
Nonalcoholic fatty liver disease (NAFLD) includes a wide spectrum of conditions from simple fatty liver steatosis to nonalcoholic steatohepatitis (NASH).
In the last decades, the rising prevalence of obesity led to a high prevalence of NAFLD during childhood, estimated at ∼7.6% in general population and 34.2% in obese children (1). Moreover, pediatric NAFLD is considered a potential independent risk factor for cardiovascular diseases.
Oxidative stress increase is considered a crucial event in the development of NAFLD, particularly in progression to NASH (8). Indeed, the increased production of reactive oxygen species (ROS) in the liver may cause lipid peroxidation, followed by inflammation and activation of stellate cells leading to fibrosis in NASH.
In this study, we provide evidence that children with biopsy-proven nonalcoholic fatty liver disease (NAFLD) have nicotinamide adenine dinucleotide phosphate (NADPH) oxidase 2 (Nox2) overactivation, which is significantly associated with oxidative stress, degree of liver damage, and endotoxemia. Our findings provide a potential role for gut-derived lipopolysaccharide in eliciting systemic Nox2 activation in NAFLD.
Among the many actors of oxidative stress, nicotinamide adenine dinucleotide phosphate (NADPH) oxidase (Nox) is the major cellular source of ROS in humans and its activation has been associated with liver damages (8). The mechanism through which Nox is activated in NAFLD is still unclear. Products of gut microbiota, such as lipopolysaccharide (LPS), may be involved in this process where the binding between LPS and Toll-like receptor 4 (TLR4) could mediate Nox isoform-2 activation in Kupffer cells and infiltrating macrophages (6).
Recently, in adults with NAFLD, we found an overactivation of Nox2 as assessed by Nox2-derivative peptide (sNox2-dp) levels and a correlation between Nox2 with serum 8-iso-prostaglandin F2α (8-iso-PGF2α), a marker of oxidative stress in vivo, and cytokeratin-18, a validated marker of hepatocyte apoptosis (7).
However, to the best of our knowledge, Nox2 activation in children with NAFLD has never been explored. To address this unknown issue, we performed a cross-sectional study to assess systemic Nox2 activation and oxidative stress in children affected by biopsy-proven NAFLD. Furthermore, we analyzed the association between Nox2 activation and plasma zonulin, a marker of intestinal permeability, LPS, and liver histological characteristics.
Results
A total of 140 children (84 females and 56 males), mean age 11 years, were included in the study; 47.8% (n = 67) children were affected by NAFLD and 52.2% (n = 72) were classified as controls.
Compared with controls, NAFLD patients had higher body mass index (BMI), alanine aminotransferase (ALT), aspartate aminotransferase (AST), serum triglycerides, sNox2-dp, LPS, zonulin, and isoprostanes, and lower high-density lipoprotein (HDL) levels (Table 1).
Clinical Characteristics in Controls and Nonalcoholic Fatty Liver Disease Patients
Italic indicates significant values.
8-iso-PGF2α, 8-iso-prostaglandin F2α; ALT, alanine aminotransferase; AST, aspartate aminotransferase; BMI, body mass index; CTRL, controls; DBP, diastolic blood pressure; EU, endotoxin units; HDL, high-density lipoprotein; LDL, low-density lipoprotein; LPS, lipopolysaccharide; NAFLD, nonalcoholic fatty liver disease; SBP, systolic blood pressure; sNox2-dp, Nox2-derivative peptide.
Bivariate analysis showed that serum sNox2-dp correlated with ALT (r = 0.39, p = 0.01), triglycerides (r = 0.38, p = 0.04), LPS (endotoxin units [EU]/mL) (r = 0.40, p = 0.03), isoprostanes (r = 0.36, p = 0.0001), and HDL cholesterol (r = −0.43, p = 0.007) (Table 2). A multiple linear regression analysis, including the variables linearly associated with the dependent variable, was performed to define the independent predictors of sNox2-dp; triglycerides (standard error [SE]: 0.07; standardized coefficient β: 0.17; p = 0.03), HDL (SE: 0.27; standardized coefficient β: −0.91; p = 0.001), homeostatic model assessment-estimated insulin resistance (HOMA-IR) (SE: 1.94; standardized coefficient β: 2.97; p = 0.003), LPS (SE: 1.41; standardized coefficient β: 2.85; p = 0.045), and isoprostanes (SE: 0.04; standardized coefficient β: 0.12; p = 0.003) were independently associated with sNox2-dp levels (R 2 = 0.42).
Bivariate Analysis Between Nox2 and Anthropometrics/Biochemical Characteristics in All the Populations
Italic indicates significant values.
After stratifying the NAFLD population according to sNox2-dp quartiles (quartile 1: 5.1 ± 2.2; quartile 2: 12.5 ± 3.19; quartile 3: 25.4 ± 6.9; and quartile 4: 64.9 ± 18.9, pg/mL), we observed that, compared with the other quartiles, patients in the highest quartile had higher ALT, triglycerides, and HOMA-IR and lower HDL cholesterol (p < 0.001) (Table 3).
Group Differences in Anthropometric and Biochemical Characteristics According to Quartiles of sNox2-dp in Nonalcoholic Fatty Liver Disease Patients
Italic indicates significant values.
HOMA-IR, homeostatic model assessment-estimated insulin resistance; UA, uric acid; WC, waist circumference.
Within the NAFLD group, compared with children without NASH, children with NASH had higher HOMA-IR, triglycerides, fasting insulin, glucose, sNox2-dp, LPS, and isoprostanes (Table 4).
Differences Between Not-Nonalcoholic Steatohepatitis and Nonalcoholic Steatohepatitis Patients for Anthropometrics and Biochemical Characteristics and Intestinal Permeability
Italic indicates significant values.
NASH, nonalcoholic steatohepatitis.
Finally, we analyzed the relationship between sNox2-dp and histological features; compared with other quartiles, the highest quartile of sNox2-dp disclosed higher steatosis, inflammation, ballooning, and NAFLD activity score (NAS) (Table 5).
Histological Features in Nonalcoholic Fatty Liver Disease Group
Italic indicates significant values.
NAS, NAFLD activity score.
Thus, as shown in Table 6, sNox2-dp levels positively correlated with steatosis (p < 0.001), inflammation (p < 0.0001), ballooning (p < 0.0001), fibrosis (p = 0.03), NAS (p = 0.0001), and LPS (p = 0.002).
Bivariate Analysis Between sNox2-dp and Histological and Laboratory Parameters in Nonalcoholic Fatty Liver Disease Group
Italic indicates significant values.
Discussion
This study provides evidence that children with NAFLD display Nox2 overactivation, associated with oxidative stress and histological grading of the liver disease.
Oxidative stress is believed to play a pivotal role in NAFLD pathogenesis and in the progression from simple steatosis to NASH (8). Increased oxidative stress in the liver gives rise to lipid peroxidation, inflammation, and activation of stellate cells leading to fibrosis in NASH (8). According to the “two hits” theory, after a “first hit” (liver steatosis related to insulin resistance, dyslipidemia, and obesity) a “second hit,” as increased oxidative stress, is required to develop NASH; however, it is still unclear why simple steatosis may progress to NASH only in some patients (8).
Nox2 is considered a main cellular source of ROS in humans and its activation has been associated with liver damages (4). In accordance with previous studies (7) performed in adults, we found increased Nox2 activity in children with NAFLD. In particular, we observed higher sNox2-dp levels in subjects with increased liver damage as those with NASH compared with those without NASH, coincidentally with overproduction of isoprostanes, a marker of oxidative stress. Furthermore, we showed a significant correlation between Nox2 activation and histological grading of liver damage such as steatosis, inflammation, ballooning, fibrosis, and NAS.
Based on this, we sought to investigate the mechanism accounting for Nox2-overactivation and focused on LPS, as previous study showed that LPS activates Nox2 via interaction with its receptor TLR4 (6). Thus, increased levels of LPS have been detected in NAFLD patients and experimental studies showed a significant association between LPS and liver damage suggesting LPS as a potential trigger of NAFLD (2). Experimental data are in favor of this hypothesis because LPS behaves as an inflammation trigger via upregulating TNF and eventually eliciting NAFLD (3). In accordance with this, TLR4 antagonists reduced TLR4-induced hepatic inflammation in obese mice via reduction of NADPH activation (3). Furthermore, Kim et al. (6) showed that TLR4 elicited Nox2 activation in hepatic macrophages and that Nox2-deficient mice were protected against high-fat diet-induced hepatic steatosis and insulin resistance. The novel finding of our report is a significant correlation between LPS and Nox2, which would reinforce the hypothesis of a potential interplay between endotoxemia and Nox2 activation but does not establish a cause–effect relationship, which should be explored by interventional trials aimed at lowering LPS.
We also investigated the mechanism accounting for enhanced LPS and focused on changes of gut permeability, which is increased in settings associated with metabolic disorders such as diabetes mellitus and obesity (5). To explore this issue, we measured serum zonulin, which modulates gut permeability by disassembling the intercellular tight junctions (9). Experimental and clinical studies demonstrated that zonulin upregulation plays a role in increasing gut permeability and that serum level of zonulin correlates with enhanced intestinal permeability (9). Furthermore, we found that zonulin and LPS levels are increased in children with NAFLD, suggesting that an increase of serum LPS levels could reflect enhanced gut permeability in patients with NAFLD.
There are some implications and limitations in our study. The study provides further insight about the role of oxidative stress on liver damage in NAFLD; in particular, we show that Nox2 is precociously activated and related to histological liver damage in children with NAFLD. Future studies must evaluate the predictive role of this marker on NAFLD development and progression. We did not evaluate other NADPH isoforms, such as Nox1 (expressed in hepatic sinusoidal endothelial cells) and Nox4 (expressed in hepatocytes, Kupffer cells, and hepatic stellate cells) that could also contribute to increased oxidative. We did not perform intestinal biopsies to evaluate the correlation between plasma levels of zonulin and integrity of tight junctions within the small intestine. Finally, the cross-sectional nature of the study does not allow establishing a cause–effect relationship among Nox2 activation, LPS and liver damage. Interventional studies aimed at lowering LPS should be undertaken to explore such hypothesis.
This study provides evidence that children with NAFLD have Nox2 overactivation, which is significantly associated with the degree of liver damage. The close relationship between Nox2 and LPS serum levels suggests a potential role for gut-derived LPS in eliciting systemic Nox2 activation.
Notes
The following children, who were referred to the “Hepatometabolic Department” of the “Bambino Gesù” Children's Hospital and to the Pediatric Clinic of “Sapienza” University of Rome from January 2014 to November 2017, were enrolled in this study and all parents provided written informed consent. Sixty-seven children, referred to the Hepatometabolic Department underwent liver biopsy for suspected NASH. During the initial investigations, liver fat was identified by abdominal ultrasound evaluation by the same experienced radiologist, who was blinded to the health conditions of the patients, using an Acuson Sequoia C512 scanner equipped with a 15L8 transducer (Universal Diagnostic Solutions, Oceanside, CA). Liver steatosis was defined according to criteria based, a bright hepatic echo pattern compared with echo response of the right kidney. Other causes of steatosis were excluded, such as viral hepatitis (A, B, C, cytomegalovirus and Epstein–Barr virus), autoimmune or metabolic liver diseases, alpha-1-antitrypsin deficiency, Wilson's disease, celiac disease, total parenteral nutrition, and use of drugs known to induce steatosis. Patients with diabetes, genetic dyslipidemia, and genetic syndromes were excluded.
Seventy-three healthy controls, matched for age and gender, were recruited at the Pediatric Clinic of “Sapienza” University of Rome. In all children, enrolled anthropometric and clinical parameters (weight, height, BMI, and waist circumference) were measured using standardized methods. Total cholesterol, low-density lipoprotein cholesterol, triglycerides, uric acid and AST and ALT, gamma-glutamyl-transpeptidase, bilirubin, albumin, and international normalized ratio were measured by standard methods.
Liver biopsy
According to the recommendation of the Hepatology Committee of the European Society of Paediatric Gastroenterology, Hepatology and Nutrition (ESPGHAN), all patients underwent liver biopsy to exclude other liver diseases, or to assess the severity of liver disease, suspected by clinical and laboratory evaluation using tests such as hypertransaminasemia and hepatosplenomegaly.
Liver biopsies were performed in all children using an automatic core biopsy device (BioPince, Amedic, Sweden) with an 18G needle that is 150 mm long. All biopsies were at least 18 mm in length and were assessed by a single liver pathologist. Biopsies were routinely processed (formalin fixed and paraffin embedded) and analyzed by different stainings. The main histological features, commonly described in NAFLD/NASH, including steatosis, inflammation portal and lobular, hepatocyte ballooning, and fibrosis, were scored according to the scoring system for NAFLD developed by the National Institutes of Health-sponsored NASH Clinical Research Network.
The steatosis was graded from 0 (<5% of hepatocytes) to 3 (steatosis >66%) and lobular inflammation from 0 (no foci) to 3 (>4 foci per 200 × field). Hepatocyte ballooning was graded from 0 (is none) to 2 (many/prominent balloon cells). The fibrosis score ranges from 0 (absence of fibrosis) to 4 (probable or definite cirrhosis). In particular, stage 1 is further subdivided into stage 1a (delicate perisinusoidal zone 3 fibrosis), 1b (dense perisinusoidal zone 3 fibrosis), and 1c (portal fibrosis only), the latter referring to the pattern of fibrosis sometimes seen in severely obese and in pediatric NAFLD patients. NAS ranges from 0 to 8, with a score of 1 or 2 corresponding to definitely not NASH, while an NAS 5–8 is suggestive of definite NASH.
The study design conformed to the ethical guidelines of the Declaration of Helsinki (as revised in Seoul, Korea, October 2008).
Enzyme-linked immunosorbent assay detection of sNox2-dp
Nox2-derived peptide, a marker of NADPH oxidase activation, was assessed in serum by the enzyme-linked immunosorbent assay (ELISA) method. The peptide was recognized by the specific monoclonal antibody against the amino acidic sequence (224–268) of the extra membrane portion of Nox2 (catalytic core of NADPH oxidase). Intra-assay and interassay coefficients of variation were 5.2% and 6%, respectively.
8-iso-PGF2α
8-iso-PGF2α levels were measured in serum by using a colorimetric assay kit (DRG International, Inc., Springfield, NJ).
Serum zonulin
Serum zonulin levels were measured using a commercial ELISA kit (Elabscience, Houston, TX). Antibody specific for zonulin has been precoated onto a microplate and 100 μL of standards and patient sera samples were added and incubated 90 min at 37°C. Then, a biotinylated detection antibody specific for zonulin and avidin-horseradish peroxidase conjugate was added to each microplate. The amount of zonulin was measured with a microplate autoreader at 450 nm. Values were expressed as ng/mL; both intra-assay and interassay coefficients of variation were within 10%.
Lipopolysaccharide
Plasma samples were thawed only once and used to perform specific sandwich ELISA to measure LPS (Hycult Biotechnology, Uden, The Netherlands). The kit has a concentration range of 0.04–10.0 EU/mL.
Statistical analyses
Statistical analyses were undertaken using STATISTICA software (version 2010, Chicago, IL). The Kolmogorov–Smirnov test was used to determine whether variables were normally distributed. Normally distributed data are described as means ± standard deviations (SDs). Between-group differences were analyzed by Kruskal–Wallis tests (for non-normally distributed data) or analysis of variance. Differences between percentages were assessed by the chi-squared test. Bivariate analysis was performed by Spearman's correlation; the variables with evidence of an association p < 0.10 were included in a multivariable linear regression using an automated procedure with forward selection. A p-value of <0.05 was considered statistically significant.
Sample size determination
In a cross-sectional study, sample size calculation was computed with respect to a two-tailed Student's t-test for independent groups, considering 11 pg/mL (δ) as difference for Nox2 levels between children with NAFLD and controls, 20 pg/mL as SD, 0.05 (α) as type I error probability, and 0.90 as power 1-β. The minimum sample size was n = 67 patients per group.