Intestinal NADPH Oxidase 2 Activity Increases in a Neonatal Rat Model of Necrotizing Enterocolitis

Necrotizing enterocolitis (NEC) is a complication of prematurity. The etiology is unknown, but is related to enteral feeding, ischemia, infection, and inflammation. Reactive oxygen species production, most notably superoxide, increases in NEC. NADPH oxidase (NOX) generates superoxide, but its activity in NEC remains unknown. We hypothesize that NOX-derived superoxide production increases in NEC. Newborn Sprague-Dawley rats were divided into control, formula-fed, formula/LPS, formula/hypoxia, and NEC (formula, hypoxia, and LPS). Intestinal homogenates were analyzed for NADPH-dependent superoxide production. Changes in superoxide levels on days 0-4 were measured. Inhibitors for nitric oxide synthase (L-NAME) and NOX2 (GP91-ds-tat) were utilized. RT-PCR for eNOS, NOX1, GP91phox expression was performed. Immunofluorescence studies estimated the co-localization of p47phox and GP91phox in control and NEC animals on D1, D2, and D4. NEC pups generated more superoxide than controls on D4, while all other groups were unchanged. NADPH-dependent superoxide production was greater in NEC on days 0, 3, and 4. GP91-ds-tat decreased superoxide production in both groups, with greater inhibition in NEC. L-NAME did not alter superoxide production. Temporally, superoxide production varied minimally in controls. In NEC, superoxide generation was decreased on day 1, but increased on days 3-4. GP91phox expression was higher in NEC on days 2 and 4. NOX1 and eNOS expression were unchanged from controls. GP91phox and p47phox had minimal co-localization in all control samples and NEC samples on D1 and D2, but had increased co-localization on D4. In conclusion, this study proves that experimentally-induced NEC increases small intestinal NOX activity. All components of NEC model are necessary for increased NOX activity. NOX2 is the major source, especially as the disease progresses.


Introduction
Necrotizing enterocolitis (NEC) is one of the most devastating diseases for premature infants. The disease involves injury and death of the small intestines. NEC causes significant neonatal morbidity and mortality, and survivors encounter many long-term sequelae [1]. Despite its high prevalence, and years of investigation, the pathogenesis remains unclear. NEC is thought to be a multifactorial disease process [2]. Enteral feeding, infection, and intestinal ischemia play vital roles in the pathogenesis of NEC [2]. Vascular dysfunction and inflammation may also contribute to the disease [3].
One factor postulated to play a key role in NEC pathogenesis is oxidative stress [3]. Superoxide (O 2 N -) is a potent source of oxidative stress. Four enzymatic sources of O 2 N -exist: mitochondria, xanthine oxidase/dehydrogenase, uncoupled nitric oxide synthase (NOS), and NADPH oxidase (NOX). Two of these pathways, NOS and NOX, require NADPH as a cofactor. Under homeostatic conditions, NOS generates nitric oxide, but in times of cellular injury, NOS can become uncoupled to generate O 2 N - [4]. Previously, this laboratory showed that uncoupled endothelial NOS (eNOS) activity produces increased O 2 N -in neonatal rat mesenteric arteries in experimentally-induced NEC [5]. However, O 2 Nproduction in the small intestines has not been thoroughly examined.
There are several NOX isoforms, which generate either O 2 N -or hydrogen peroxide (H 2 O 2 ) [6]. NOX enzymes contribute to oxidative stress in both physiologic and pathologic states. The first two isoforms (NOX1 and NOX2) generate O 2 N -. NOX2, found predominantly in neutrophils and macrophages, is the prototypical NOX isoform, and contributes to the phagocytic respiratory burst in host defense [6]. Absent NOX2 subunits results in the immunodeficiency disorder known as chronic granulomatous disease [7]. Conversely, excess O 2 N production from NOX2 occurs in many pathologic states, including septic shock, cardiovascular disease, and diabetes [6,[8][9][10][11][12].
NOX1 also generates O 2 N -, but has a more diffuse distribution [6]. The physiologic functions of NOX1 are incompletely understood. Vascular remodeling and angiogenesis are associated with NOX1 [13,14]. As with NOX2, elevated O 2 Nproduction from NOX1 may contribute to inflammation and injury. Elevated NOX1 activity is associated with vascular injury [15] and preeclampsia [16]. In addition, gastrointestinal diseases such as small intestinal mucositis [17], inflammatory bowel disease [18], and colon cancer [19] have been associated with increased NOX1 levels.
Inflammatory processes often involve leukocyte activation. Previously, leukocytes have been shown to be mediators of injury in experimental NEC animal models [20]. The etiology for the injury is not well understood. Because O 2 N -generated by NOX2 increases in leukocyte activation, the association between O 2 N -levels and NEC needs further understanding. The contribution of NOX to the overall O 2 N -generation in NEC and its relevance to it pathogenesis has not been studied. The Sprague Dawley newborn rat model of NEC has been extensively utilized in basic science research for NEC, and provides a comparable phenotype to that observed clinically. We hypothesize that the generation of O 2 N -by NOX1 and NOX2 play crucial roles underlying the inflammation of NEC. In this study, we examine how experimentally-induced NEC is associated with NOX-dependent O 2 N -production. In addition, the contribution of specific NOX isoforms will be delineated.

Results
Experimentally-induced NEC is associated with increased NAPDH-dependent O 2

N -production
In order to confirm that small intestinal O 2 N -generation is via NOX and NOS, samples were analyzed in the presence or absence of NADPH as described above. When NADPH was withheld from the reaction, there was an absence of chemiluminescence in every sample.
After confirmation that O 2 N -generation is NADPH dependent, control and NEC samples from D4 were analyzed (Fig. 1). There was a three-fold increase in O 2 N -production in NEC animals compared to controls (p,0.05).
LPS, Hypoxia, and Formula are required for increased O 2 N generation Because pups exposed to all three components of NEC had increased O 2 Ngeneration, it was important to determine the contribution of each component in causing elevated O 2 N -production. Accordingly, the study was repeated with pups received formula (F), formula and hypoxia (F/H), and formula and LPS (F/LPS) (Fig. 2). Pups who received F, F/H, or F/LPS did not have increased NOX activity compared to control pups (p5 NS by ANOVA). However, NEC pups had increased activity compared to pups in any other condition (p,0.05 for each group compared to NEC or ANOVA p,0.0001 with NEC group).

N -exhibits temporal differences
After confirmation that all components of the NEC model were required for maximal NOX activity, the assay was used to quantify O 2 N -production over the first 5 days of life (Fig. 3). In control rats, there were no significant changes in O 2 N -production from day of life 0 (D0) to D1, nor D1 to D2. Pups from D3 were significantly decreased compared to D2 (p,0.05). From D3 to D4, there was another significant increase (p,0.01). Comparing D0 to D4, there was nearly a three-fold increase NOX activity (p,0.05).
NEC pups also exhibited changes during the study. There was a significant decrease from D0 to D1 (p,0.005). On D2, O 2 N -production significantly increased over D1 (p,0.05). D3 samples exhibited a non-significant increase from D2. A four-fold increase, the largest in the study, occurred on D4 (p,0.01). From D0 to D4, there was also a four-fold increase in activity (p,0.05).
In addition to changes observed within each condition, control and NEC pups exhibited different activities on the same day of the study. NEC pups had significantly elevated O 2 N -generation activity when compared to controls on D0, D3, and D4 (p,0.05).
NOX2, but not NOS, is a significant source of O 2

N -generation
Because NOS and NOX can generate O 2 N -and require NADPH, additional studies were needed to elucidate specific enzymatic contribution. The chemiluminescent assay was performed on tissue homogenates in the presence of specific inhibitors for NOS (N v -Nitro-L-arginine methyl ester hydrochloride, L-NAME) and NOX2  (GP91-ds-tat). Control and NEC samples from D4 were exposed to inhibitors, and O 2 N -production quantified (Fig. 4). Although L-NAME tended to reduce O 2 N -production in both control and NEC samples, the differences did not achieve significance (p.0.05).
In contrast, GP91-ds-tat reduced O 2 N -production in both control and NEC samples by nearly 90% (p50.01). Importantly, samples from NEC pups treated with GP91-ds-tat generated O 2 N -at essentially the same levels as to untreated control samples.

GP91 phox expression is increased in NEC
To further examine specific enzyme contributions, RT-PCR was performed for GP91 phox (NOX2 catalytic site), NOX1, and eNOS. Results were normalized to expression observed on D0 from control pups. GP91 phox expression in control pups exhibited minimal variation among the five days studied (Fig. 5). Differences observed were not significant (p.0.1 by ANOVA). The highest expression occurred on D3 (1.47¡0.18 fold increase over D0). There were no significant differences observed between days.
Similar to the chemiluminescent study, same-day differences were observed between groups. Expression was significantly decreased in NEC pups on D0 (1¡0.096 vs. 0.6¡0.08 fold, p,0.01). On D2, expression in NEC pups was threefold greater than controls (4.54¡0.8 vs. 1.28¡0.16, p,0.005). Expression in NEC animals was also significantly increased compared to control samples on D4 N -production in small intestinal homogenates from D0-4. Same-condition experimental groups that had significant differences represented by error bars. Both control and NEC samples were significantly different between D0 and D4. In addition, significant differences between control and NEC samples occurred on D0, D3, and D4 (p,0.05). n55-6 per group * 5 p,0.05, ** p,0.01.

Expressions of NOX1 and eNOS do not increase in NEC samples
Because NOX1 also can generate O 2 N -, RT-PCR for this isoform was measured (Fig. 6). Both control and NEC samples exhibited significant differences during the study (p,0.05 by ANOVA for each condition). However, different trends were observed in the two conditions.
NEC pups had different expression patterns in NOX1. Expression increased from D0 to D1 (0.56¡0.099 vs. 1.51¡0.34, p,0.05), followed by a decrease from D1 to D2 (0.62¡0.11, p,0.05). While NOX1 expression increased on both D3  Previous research from our laboratory identified uncoupled eNOS as a source of O 2 N -in NEC in isolated mesentery [5]. To determine if eNOS is involved, RT-PCR for eNOS was performed (Fig. 7). In contrast to GP91 phox and NOX1 expression, values for eNOS were not significantly different over the time course for either control or NEC samples (p. NOX2 subunits co-localize on D4 of NEC There are six subunits that comprise the NOX2 complex: two transmembrane proteins (GP91 phox and p22 phox ) and four cytosolic proteins (p40 phox , p47 phox , p67 phox , and Rac). In the inactive state, the cytosolic proteins do not interact with the transmembrane proteins. When p47 phox is phosphorylated, the cytosolic proteins translocate to the GP91 phox /p22 phox complex. This collection of proteins is now capable of generating superoxide by reducing molecular oxygen with NADPH as a cofactor [21]. It is possible to indirectly assess the degree of NOX activity with studies that examine how specific subunits co-localize. Previous studies have demonstrated that co-localization of p47 phox and GP91 phox can be observed with immunofluorescence [22,23]. To add evidence that neonatal pups have increased NOX2 activity after intestinal injury in this NEC model, we performed immunofluorescence for p47 phox and GP91 phox .
Representative samples from D1, D2, and D4 are seen in Fig. 8. The terminal ileum was evaluated by immunofluorescence for GP91 phox (green) and p47 phox (red), and for co-localization of the two proteins (yellow) as an index of activation. In the control terminal ileum (Fig. 8a), GP91 phox and p47 phox colocalization was minimal, and not increased compared to any of the control samples. In NEC terminal ileum (Fig. 8b), there was minimal co-localization of GP91 phox and p47 phox on D1 and D2. However, co-localization of GP91 phox and p47 phox on D4 was increased compared to all other samples.

Discussion
NEC remains one of the most significant complications of extreme prematurity. Despite a high prevalence with serious morbidity and mortality, little improvement has been achieved towards understanding this disease. Inflammation and oxidative stress have been demonstrated to be integral to the development of NEC. Several studies have shown that premature infants with higher levels of oxidative stress markers are more likely to develop NEC [24][25][26]. In animal models of NEC, elevated levels of ROS have been implicated in the pathogenesis of NEC [27][28][29][30]. However, the enzymatic sources of increased ROS have not been thoroughly examined. NOX enzymes are significant sources of ROS in many diseases such as hypertension, diabetes, and atherosclerosis [31]. This study is the first to show that NOX-dependent O 2 N -production is increased in an animal model of NEC.
Several changes occurred in the NOX activity in this NEC model. There was nearly a five-fold difference in NOX activity in pups exposed to the combination of formula, hypoxia, and LPS compared to breast fed pups. Our data also indicate that all three components of the NEC model (formula feeding, hypoxia, and enteral LPS) are required to cause elevated NOX activity. Subjecting pups to only one or two stressors did not change NOX activity. This is consistent with clinical NEC, which is considered a multifactorial disease.
Uncoupled NOS has the potential to generate O 2 N -and requires NADPH as a cofactor. We examined whether iNOS or eNOS could be responsible for the changes observed. Expression of eNOS did not vary significantly in either control or NEC pups during the course of the study. We did not measure iNOS mRNA, as it has been shown to increase in experimental NEC [32][33][34]. L-NAME inhibits all forms of NOS [35,36]. When L-NAME was included in the reaction, O 2 Nproduction was unchanged, indicating that both eNOS and iNOS are not significant sources of O 2 N -in this model. Based on RT-PCR and inhibitor studies, NOX2 appears to be the most significant isoform responsible for changes in O 2 N -production. NOX2 mRNA expression increased dramatically on D2, compared to both same-day control and previous day NEC samples. This changed preceded the largest increase in O 2 Nproduction. Conversely NOX1 mRNA did not change as dramatically. At every Fig. 8. Immunofluorescence for p47 phox and GP91 phox on D1, D2, and D4. Fig. 8a shows control samples, and Fig. 8b shows NEC samples. The images are of p47 phox (red) GP91 phox (green), and the co-localization of the two proteins (yellow). N -production was decreased by 90% in the presence of GP91-ds-tat in both control and NEC pups. This indicates that NOX2, and not NOX1, is the main source of O 2 N -in this model. The immunofluorescence studies also provide evidence that NOX2 is an important source of O 2 N -generation in this model. Samples from control pups demonstrated relatively minimal co-localization of GP91 phox and p47 phox on days examined, indicating that p47 phox is most likely not activating NOX2 activity in healthy newborn intestines. NEC samples also had little co-localization on D1 and D2. However, there were dramatic increases in GP91 phox /p47 phox co-localization on D4 in NEC pups. This indicates that GP91 phox and p47 phox are not interacting in control samples or in the first few days of exposure to intestinal injury. However, p47 phox localizes to GP91 phox on D4, when O 2 N -production was the highest. This indicates that NOX2 activity is increased, and a major source of O 2 N in this model.
Prior studies have shown important post-natal changes that are altered in NEC such as barrier function and LPS detoxification [37,38]. Similarly, we demonstrate that changes in NOX activity occurred with postnatal age. Pups delivered one day prematurely via cesarean delivery had higher NOX activity at birth. This increase in NOX activity could be from prematurity or not undergoing labor, and ultimately requires further investigation. NOX activity rose among control pups during the five days of the study. The increases were significantly different when comparing D0 and D4, and most likely represent physiologic intestinal maturation. In NEC pups, there were two notable changes. The first was from D0-1, when NOX activity decreased. The second and most dramatic differences occurred from D2-3 and D3-4. NOX activity increased several fold during this time. These results imply that repeated exposure to injurious stimuli elevates NOX activity.
These results contribute to the understanding of NEC in several manners. First, this is the first study to associate increased NOX2-dependent O 2 N -in NEC. While oxidative stress has been suspected in the disease, our study identifies changes in a known reactive oxygen species source. We also demonstrate that NOX1 and eNOS are not important sources of O 2 N -in this model. Another important finding is that NOX activity increases only when the combination of formula feeding, hypoxia, and enteral LPS are present, which demonstrates that multiple factors are required to induce injury in this model, and further underscores the complicated nature of NEC.
There are some limitations in this study. At the time of this study, there was not a NOX2 deficient rat strain. GP91-ds-tat was initially designed to prevent the binding of p47 phox to GP91 phox , but may also prevent p47 phox from activating NOX1 [39,40]. However, no studies have clearly demonstrated that NOX1 activity is reduced by GP91-ds-tat. NOX1 activation is more efficient with the NOXO1 and NOXA1 subunits than the NOX2 subunits, so the degree of NOX1dependent O 2 N -secondary to p47 phox activation is probably minimal. The immunofluorescence studies do not prove that NOX2 is activated on D4 in NEC animals. However, the dramatic differences observed on D4 compared to other time points and experimental conditions indicate that NOX2 activity is elevated. In closing, this study provides novel evidence that NOX2-dependent O 2 Nincreases in a rat model of NEC. The results provide evidence that oxidative stress is an important factor in the development and progression of NEC. Inhibition of NOX2 may reduce the incidence and severity of NEC. Future studies will involve the use of genetically modified animals that lack NOX1 or NOX2 to confirm findings from this study.

Ethics Statement
All animal experiments for this study were carried out in strict accordance with the recommendations and approval from the Institutional Animal Care and Use Committee (IACUC) at the Medical College of Wisconsin (Animal Use Application #0092). All members of the research team received appropriate training and were in good standing with the IACUC.

Animal Model
An in vivo NEC model was utilized as previously described [5]. Pregnant female Sprague-Dawley rats (Harlan Laboratories, Madison, WI) were obtained prior to their estimated date of delivery. Control pups were full-term Sprague-Dawley rat pups spontaneously delivered and dam-fed. Experimental pups (NEC pups) were delivered one day prematurely by cesarean section. NEC pups were maintained in an incubator at 37˚C and orogastrically fed 0.1-0.25 ml of Esbilac formula, supplemented with Neocate Powdered Formula (Abbott Laboratories, Abbott Park IL) containing LPS (2 mg/kg/day, Sigma-Aldrich, St. Louis, MO), thrice daily when the pups were D0-3. After each feeding, NEC pups were subjected to hypoxia (FiO 2 of 5% via a hypoxia chamber) for 10 min. Some pups were subjected to only parts of the NEC injury model (formula, formula/hypoxia, or formula/LPS) to clarify contributions from each component. Pups were euthanized on each day of life (D0-D4) to establish a time-sensitive pattern. Pups were euthanized prior to feeding with a ketamine/xylazine injection as per internal IACUC guidelines. One segment was placed in RNAlater (Ambion, Grand Island, NY) and stored at 220˚C until isolation. The remaining small intestine was analyzed for NADPH dependent O 2 N -production.

N -Production
Intestinal segments were placed in 500 ml of a sucrose-potassium phosphate buffer. Segments were homogenized at 4˚C, centrifuged at 13,000 rpm at 4˚C for 10 min to remove cell debris, and the supernatant recovered. Protein content was quantified with Bicinconinic acid (BCA) reagent per the manufacturer's instructions (Pierce Chemical Company, Rockford, IL). Production of O 2 N -in small intestines homogenates was quantified as Tironinhibitable lucigenin chemiluminescence. Aliquots of 50-300 mg of homogenate were added to wells containing the following reaction components (final concentration): NADPH (100 mM), sucrose (150 mM), and lucigenin (5 mM) (all chemicals from Sigma-Aldrich) [41,42]. To confirm that NADPH was necessary for O 2 N -generation, the cofactor was withheld from the reaction mixture for each sample. Chemiluminescent signals were collected over 30 minutes on a 96 well plate (Bio-Rad), and the total RLU summated (luminometer from Turner Scientific, Madison, WI). Samples from control and NEC groups were analyzed from D0-4. On D4, inhibitors were added to the reaction mixture to quantify contribution from specific enzymes. GP91-ds-tat (50 mM, Blood Center of Wisconsin, as previously described [40]) was used to determine the contribution of NOX2 to O 2 N -production in homogenates. L-NAME (500 mM, Sigma-Aldrich) was added to determine the contribution of O 2 N -production from all NOS isoforms. Background signals were determined by performing the reaction with the respective inhibitors in the absence of an intestinal homogenate, and the resulting chemiluminescence was subtracted from the chemiluminescence observed. The average signal in the presence of Tiron (Sigma-Aldrich) was calculated and subtracted from each sample. Superoxide anion results are expressed as RLU) inhibited by Tiron normalized to intestinal protein (RLU/mg).

Real-Time PCR (RT-PCR)
RT-PCR for the membrane-bound catalytic site for NOX2, GP91 phox (Refseq Accession Number NM_023965.1, primer from Superarray Biosciences, Valencia, CA), NOX1 (Refseq Accession Number NM_053683, primer from IDT, Coralville, IA), and eNOS (Refseq Accession Number NM_021838.2, primer from IDT) were used to quantify time-dependent changes in expression in the terminal ileum. RT-PCR for GAPDH (Refseq Accession Number NM_017008.3, primer from IDT) was used as an internal control. Total RNA was isolated using the Qiagen RNeasyMini Kit, per the manufacturer's protocol. RNA concentration and purity were determined on a NanoDrop spectrophotometer (Molecular Devices, Sunnydale, CA). RNA integrity was measured using the BioRad Experion Analysis Kit; samples with an integrity value of less than 9.0 were excluded (3/53 samples excluded). Complementary DNA (cDNA) was synthesized from 2 mg of DNasetreated total RNA using the iScript cDNA Synthesis kit (Bio-Rad). RT-PCR was performed using ABI Prism 7900HT software (Applied Biosystems, Carlsbad, CA) together with SYBR Green RT-PCR Master Mix (BioRad). All gene-amplification reactions were performed in triplicate. Expression was calculated using the Pfaffl method to determine the fold change in expression in NEC animals relative to control samples from D0 [43].

Immunofluorescence
Following euthanasia, a 1 cm segment of terminal ileum was embedded in paraffin and cut into 4 micron-thick sections. The sections were deparaffinized at 56˚C, immersed in xylene three times and hydrated with ethanol (two times with 100%, two times with 95% and one time with 75% ethanol) for 5 min. For antigen unmasking, slides were heated in 10 mM sodium citrate buffer (pH 6.0) for 15 min prior to treatment with 0.3% hydrogen peroxide for 30 min. The specimens were treated with 5% BSA in TBS-T for 30 min at room temperature followed by overnight incubation with the following primary antibodies, mouse anti-GP91phox, anti-rabbit p-47phox at 4˚C, followed with goat anti-rabbit DyLight-649 and goat anti-mouse 488 conjugated 2nd antibody. Images were acquired by confocal microscopy using a Zeiss LSM510.

Statistical Analysis
A one-way ANOVA was used to determine differences among samples from the same condition for the chemiluminescent experiments. RT-PCR results were calculated from the Pfaffl method, and normalized to control samples from D0 for each respective enzyme. An unpaired t-test was used to compare two specific results for both the chemiluminescent and RT-PCR experiments. Variation was expressed as standard error of the mean (SEM).